Apple's "Time Machine" Now For Linux... Sort Of

deander2 writes "Apple's 'Time Machine' is cool, but I use Linux, not MacOSX. So here is a Linux implementation (built off of rsync, of course). No fancy OpenGL, but quite functional none-the-less."

Comments

Question

[stoolpigeon]

I have not used Leopard - so this is a real question, not a snarky response. My understanding was that a large part of what makes the whole Time Machine work and worthwhile is the interface. So if you don't have that, isn't it just another backup tool? Let me reiterate - this isn't a rhetorical question. Is doing the same thing without the interface sufficient or is it missing the point?

Re:Question

[Anonymous Coward]

RTFA. (I know, I know.) This is actually just a GUI that brings together existing Linux technologies to provide a Time Machine type of system. So in that respect, it's almost exactly like Time Machine. Whether it's truly as easy to use as Time Machine or not is something only the users of Flyback can answer.

Re:Question

[krog]

Another thing which makes Time Machine so cool is that it is hooked into the filesystem at a low level. Rather than having to inspect the entire directory tree rsync-style, Time Machine uses the FSEvents interface to stay informed of filesystem changes. FSEvents isn't perfect (it actually only records when a directory's contents have changed) but it beats rsync-ish traversal any day.

In my opinion, without such a method for watching FS changes as they occur (or later, from a log), any hackish solution will fall far short of Time Machine's performance.

Re:Question

[dlsmith]

I don't know whether this Linux implementation does something like it, but what I like most about Time Machine isn't the interface. It's the fact that the backup utility takes care of disk management automatically.

My current backup strategy works something like this:

1. Set up Retrospect nightly backup scripts.
2. Happily enjoy the security of having backups for a few weeks.
3. Wake up in the morning to see an "external disk full" error message.
4. Procrastinate for weeks while I try to decide whether I'd rather trash the entire archive or find someplace to dump my 80 GB of data (which probably involves making space somewhere, which is always a project).
5. Finally get fed up with having no backups and just discard the archive.
6. Return to step 1.

If I were smart and vigilant, I would catch when the archive reaches about 30 GB, and create a new one then, so that managing older archives could be done in more tractable chunks. If I were rich, I would just buy a number of external drives that I would rotate as they filled up. But I am apparently neither, so I just get stuck in this cycle in which I only have a current backup 1/3 of the time, and older archives are randomly discarded or distributed wherever I can find the space.

The great thing about Time Machine is that it consistently fills up my disk with the most relevant backup data: current backups at a high frequency, and months-old backups at a low frequency. When space runs out, the oldest data gets thrown away, but the quantum chunk is a diff between backups, not an entire 80 GB archive.

Re:Question

[Brandon30X]

http://en.wikipedia.org/wiki/Inotify

And before that, now obsolete...

http://en.wikipedia.org/wiki/Dnotify

-Brandon

Re:Question

[herve_masson]

Whatever the merit of this python gui, comparing it to time machine is far fetched, to say the least...

I think you're right: the value of time machine lies in its GUI. Much more than in its underlying file copy techniques. Like in any serious backup tool, the interface is _the_ key element. Obviously, data needs to be saved reliably somewhere, but that's something we can do in various ways for a long time.

An efficient backup/restore GUI is hard to do. This is what Apple has done beautifuly here, and this is what bring the tool in the average user's hands. This is new.

Re:Question

[morgan_greywolf]

Isn't that what ReiserFS plugins are for? Speaking of reiserfs, namesys.com seems to be down, otherwise, I'd like to the document.

Re:Question

[Khazunga]

Actually, I have mixed feelings about having a daemon following inotify (fsevents equivalent for linux) in order to backup. My setup uses backuppc, which daily rsyncs my disk and backs it up using much the same archival solution that Time Machine uses. The rsync is non-noticeable (and, in my case occurs during working hours). An inotify daemon, on the other hand, could be responding to lots of small requests that produce null results (temp files, disk writes over the same sectors, etc).

Fine-grained backups may be interesting, but I wouldn't be interested in any kind of performance drag because of it. Daily backups have served me just fine, thanks.

Re:Question

[dave420]

It's usefulness is a mixture of backing up and recovery. If you look at (and this is going to hurt to say it) Windows' Volume Shadow Copy, which does essentially the same thing (stores snapshots), to recover a previous version, you simply right-click a drive, folder or file, select "restore previous version", click a date, and it's done. Or you can restore the files to a different location so you can check them out/compare them, etc. So a decent snapshot-based system doesn't revolve around flying windows or OpenGL, just a sensible way to put data in and to get data out. :)

Re:Question

[DHalcyon]

The tiny difference here is that apples FSEvents is basically a daemon following the fsevents device (OS X equivalent of inotify) all the time keeping a log (directory level, as to not bog everything up with a lot of small changes), so the backup app does not have to run in the background all the time. It also notices when the disk changed while not mounted in OS X (In which case it only tells you _that_ the disk changed, I believe. You still have to figure out what changed, exactly yourself).

Re:Question

[Em Adespoton]

...enter BackupPC. Doesn't work at the FS level, so it's not Time Machine, but you can set it to do daily backups to cut down on the number of traversals needed. Has the added benefit of working over smb, rsync, and/or ssh. Also scales to large collections of PCs, and works better in a multi-PC office. PLUS, it can back up any device that can use smb, rsync, or ssh.

Re:Question

[Anonymous Coward]

Time Machine does _not_ use FSEvent, in ordinary Apple fashion, it uses an undocumented internal API that exposes more information than FSEvent.

Re:Question

[krog]

Unless you can produce a link or some debugger output or something, I'm gonna go ahead and trust ArsTechnica more than I trust you.

Re:Question

[Domini]

The great thing about Time Machine is you just plug in a hard drive and go!

It's got APIs which programs could use to represent content-sensitive revision-based restores... such as mail message, photos etc.

The fact that it uses unix-like hard symlinks in the background is just a good design.

It's also got a brilliant (and cool) UI.

Sorry... rsync it just well... rsync.

Re:Question

[MasterVidBoi]

You're exactly right. It is missing the point without the interface. The real breakthrough in TM's UI isn't that the user can go retrieve a file as it was last week (with gratuitous 3D effects), but rather that they can go retrieve something that isn't a file.

What if you deleted that email you really wanted, or made a bad edit to a contact in your address book, or a photo in iPhoto/Picasa? These apps store lots of data in some kind of database. As a geek, you know that you need to find this database, move the current one aside, restore the old one, export the content you want from the app, move the current database back into place, and import the content you just extracted from the old database.

With TM, Your Mom opens Mail, and presses the TM button. She gets the same 'windows through time' view, listing her mailbox at each checkpoint. She selects the message(s), and hits the restore button, and it gets brought into the current database. She doesn't care how it gets represented on disk.

See this screenshot: The user isn't browsing files, they are browsing contacts: http://scrap.dasgenie.com/images/017-TimeMachine.png

TM is implemented as file-based backup (with a few less common twists), but that isn't how the UI presented to the user. Without the UI, it's Yet Another Backup Solution.

Re:Question

[alanQuatermain]

FSEvents was written primarily for the use of the time machine backupd process, which runs via launchd as a cron-like job every hour. They actually wrote FSEvents because, unlike the Spotlight daemon (which reads directly from /dev/fsevents) they didn't want the backup app to run constantly. FSEvents runs a constant background daemon for you, which simply builds up an aggregate log of everything it reads from the device.

Re:Question

[jwthompson2]

I'm using Time Machine and the ease of setup is great for the non-tech savvy (System Preferences > Time Machine > On/Off and then pick a disk). The interface is what makes it more than just another backup tool, the way you can browse backup sets is unique as far as I'm aware. I'm accustomed to automated backup utilities and on that front Time Machine isn't anything special. The way you can work with the backup sets is really what makes it useful to me.

Re:Question

[dgatwood]

AFAIK, if it changed on another Leopard machine, you should get a complete list of changes. You should only get a stream reset if it was modified on a non-Leopard machine (or if you're working with a non-HFS volume).

Re:Question

[samkass]

The thing that makes Time Machine innovative is how information is retrieved, and the thing that enables that is the manner in which it's set up, the technologies it uses, and the APIs it offers.

The things Time Machine does especially well (the combination of which doesn't exist in any other product I'm aware of):
* Do everything automatically, from handling the availability (or lack thereof) of the backup volume to configuring the backup system. It sets up incremental backups so you can restore something as of every hour for the last 24 hours, every day for the last month, and every month for the last year.
* Track changes in the filesystem without having to poll, diff or stat modified times for every file (using MacOS X's built-in support for this feature). Thus, it doesn't thrash a drive looking for changes every hour but still gets everything.
* Allow retrieval from within an application UI, so I don't have to leave Address Book, iPhoto, Finder, or whatever to use it.
* All the end-user to easily query back through time. Type in a search string (within the application's own search UI) and zip back in time to the last time the results of that query change-- this in particular is something I'm not aware that any other backup solutions has ever done.

Re:Question

[Anonymous Coward]

You're confusing FSEventsd with /dev/fsevents. The way in 10.4 to monitor file system events was to feed directly from /dev/fsevents. In Leopard, FSEventsd now serves out relevant info from /dev/fsevents into a log file in order to keep an errant process from holding up /dev/fsevents. The /dev/fsevents is now more for trusted recipients of raw FS event data; things tied closer to the core of the OS, like Time Machine and Spotlight. But it isn't undocumented, just discouraged now that you can just read a log file.

Re:Question

[empaler]

The interface is indeed the entire cake; it also seems to be what they're aiming at. I'd be more than satisfied with half the user friendliness of Time Machine.

Re:Question

[dan the person]

Is this different to standard incremental backups?

You do a full backup on monday, incrementals all week, then a full backup next monday. You keep one full backup for each week of the last month, and the first backup of the month for previous months.

Re:Question

[shmlco]

Leopard feels faster than Tiger in nearly every aspect, so there's little "drag". Further, it avoids backing up "junk" files and the way it handles "small" file changes is unique. You really should read the Ars article before passing judgement on technology you don't understand.

And the "hourly" backup has already served me quite well on at least two occasions thus far.

Re:Question

[Anonymous Coward]

FSEvents logs to a file, and then TimeMachine is run via launchd periodically ( cron ). There is minimal performance hit until the actual TM process fires.

Re:Question

[noidentity]

An inotify daemon, on the other hand, could be responding to lots of small requests that produce null results (temp files, disk writes over the same sectors, etc).

Nope, I read in the recent Ars article that the FSEvents mechanism logs events to a file, so it doesn't require each watching program to be constantly running. So I'd expect Time Machine to be active only when backing up or restoring. The log apparently stores just the directory the changed file was in, reducing the number of events in it.

Re:Question

[dlsmith]

The main problem I have with doing something like that is limited disk space. If my original content uses x bytes, you're talking about requiring a minimum of 9x bytes to follow a scheme like this for 6 months. With Time Machine there's no redundancy (both a strength and a weakness, I suppose), so, depending on how much changes between snapshots, I might be able to get away with 3x bytes. (I have no idea how accurate that number is, since I haven't actually used the thing...)

Re:Question

[TheNetAvenger]

Another thing which makes Time Machine so cool is that it is hooked into the filesystem at a low level. Rather than having to inspect the entire directory tree rsync-style, Time Machine uses the FSEvents interface to stay informed of filesystem changes. FSEvents isn't perfect (it actually only records when a directory's contents have changed) but it beats rsync-ish traversal any day.

In my opinion, without such a method for watching FS changes as they occur (or later, from a log), any hackish solution will fall far short of Time Machine's performance.

Um, this is splitting hairs a bit.

Using your logic as a basis of 'better' someone could also 'one up' Time Machine by pointing out that Vista's equivalent feature works at both the monitoring level and the 'copy-on-write' level of the FS, which is something HFS+ isn't even capable of doing.

Then point out that this is why Time Machine is slower than 'Previous Versions' on Vista and why Time Machine cannot use volume level snapshots and how Vista is keeping more granual versioning with virtually NO performance loss.

This could even be extended to why Time Machine can't use compressed backup stores to save space because of its ties to using HardLinks, and having no FS level compression support.

(Little things like this are important, ask all the Time Machine users screaming in the Apple forums as fast as Apple can delete their posts. Two patches so far, and they are still having problems because of Time Machine's need for HFS+ on the external drive. - Which also makes the drive worthless for anything but just being a Time Machine stooge.)

Bottom line:

The Linux stuff is a step in the right direction, I wouldn't start picking at details to bolster OS X, as Vista can technically and easily trump what OS X is doing as well, so back off Linux nitpicking, and encourage what is being done to make backups easier...

Re:Question

[r_jensen11]

So TimeMachine doesn't have a filesize limit? I suppose you could work around this by creating a partition just for your timemachine files, but that would probably require too much work for you too, since you havn't done it yet.

Re:Question

[Ba3r]

Don't forget the use of the extensive preview framework, enabling time machine to quickly render an accurate thumbnail based upon whatever is registered in preview for that file type. Even with a fast hackish solution, without a consistent, quick, and extensible thumbnail generation framework, that magic pixie-dustness of timeline becomes a boring, less than usable, badly managed repository.

Re:Question

[MikeFM]

IMHO there shouldn't really need to be an interface. I've been using file system versioning in Linux for a long time (thanks to FUSE) and it works just fine using normal file dialogs and explorers. I don't see an argument for a pretty interface. A minor upgrade of file dialogs and explorers would be plenty to make the extended functionality seamless but it's not a big deal since it's easy to work with old versions as they are just part of the normal filesystem.

Re:Question

[sciurus0]

The answer is to set a backup delay, for example 5 minutes. When a file is changed, the delay counter begins. If the file changes again before the delay is over, the counter is reset. If the file still exists when the counter is over, it's backed up. This is what TimeVault does. Like flyback, TimeVault is written in python and GPL-licensed. Unlike flyback, Timevault:

• Uses inotify
• Tracks backups in a sqlite database
• Separates the backup/restore code into a daemon
• Has a dbus interface
• Has nautilus and system tray integration

Also, the author is looking into storing just the parts of files that change (i.e. like rdiff-backup instead of rsnapshot/dirvish). Thanks to the separation of concerns and dbus interface, I think this program could become a standard component of the free software desktop.

Re:Question

[aduzik]

The implementation is also so cool. Apple added a new API called FSEvents (file system events) that notifies Time Machine whenever any file changes. That way, Time Machine always knows what files to back up. They also augmented the file system to allow "multi-hard links". So, think of a hard link in Linux, then imagine adding a reference-counting scheme to it so every file knows how many hard links it has, then imagine adding the ability to hard link entire directories, too. All of this is part of the POSIX specification, but hardly anyone ever implements it. But it does allow for space-efficient incremental backups without resorting to any additional jiggerypokery in Time Machine to make it all work. For the most part, it's difficult to implement and not very useful outside of an application like Time Machine.

AppleInsider did a much better job than me of explaining all of this in their "Road to Leopard" series. Page 2 gives the technical details on Time Machine: http://www.appleinsider.com/articles/07/10/12/road_to_mac_os_x_leopard_time_machine.html&page=2.

I'm too lazy to do any research...

[snark23]

...but how is this different Dirvish, which has been around for years?

Re:I'm too lazy to do any research...

[snark23]

To answer my own question, there are apparently two differences:

1. The technical difference: Apple's "Time Machine" is implmented at the file system level via ZFS. Dirvish's integration is less tight, allowing you to use whatever file system you want, but requiring the historical/snapshot data to be stored in a specially set-aside directory.

2. The non-technical difference: "Time Machine" has a sexy user interface. Dirvish does not. It's easy for CLI nerds to dismiss this point, but by reading the comments here, it's apparent that people really like this sexy user interface. "Flyback" is mostly trying to duplicate that interface.

Re:I'm too lazy to do any research...

[Anonymous Coward]

time machine does not use ZFS. In fact, Leopard as shipped cannot write to ZFS partitions. ZFS would be the "correct" way to handle it, there current implementation is a hack, as this rsync implementation shows. It's the type of solution you'd expect from a project in an Advanced OS course.

Yes, they have a nice sexy inteface (and that's good and all), but in truth there interface isn't really that amazing. For most people, microsoft shadow copy interface is just as usable.

Re:I'm too lazy to do any research...

[ShieldW0lf]

Just to clear this up for everyone:

Computers do not have sexy interfaces. Neither do phones or mp3 players.

A sexy interface is pink, and moist, and warm, and throbs gently. It occasionally releases strange sounds, odd odors and fluids of various colors. It comes with no manual to operate, and apparently many people never figure them out properly, however, it is a great deal of fun to experiment with them.

I absolutely love sexy interfaces, and I'm going to go enjoy one right now.

If you'd like a sexy interface of your own, here's a hint. They doesn't come from Apple.

Re:I'm too lazy to do any research...

[CarpetShark]

but how is this different Dirvish...?

It's almost as good as faubackup? ;)

Re:I'm too lazy to do any research...

[Anonymous Coward]

Dirvish looks nice, but has the same problem so many cool Linux tools have. It stops short of being a backup "for the rest of us". An approachable GUI is important to most computer users as well as a "set it and forget" it maintenance cycle.

However, most importantly hard links on directories were added to the OS so that entire unchanged directory trees would not be reproduced. This significantly reduces the number of files needed on the backup drive.

Hard to beat the Time Machine setup scenario: 1. Click the big ON button; 2. Pick a disk drive
Done :)

Re:I'm too lazy to do any research...

[Anonymous Coward]

Apple's "Time Machine" is implmented at the file system level via ZFS.

Bullshit. If you don't know what the hell you're talking about, shut the hell up.

Re:I'm too lazy to do any research...

[StarfishOne]

You forgot the magic pair of words: plug and play

Re:I'm too lazy to do any research...

[Anonymous Coward]

For most people, microsoft shadow copy interface is just as usable.

Usable?! How many people even *know* about Shadow Copy?

Apple stick a freaking Time Machine icon on the dock and filled the internet with details about it. Is there a single geek or apple leopard user that doesn't know what Time Machine is?

The first step towards usability is making it available to everybody, automatically, and in a non-scary and accessible manner. Apple have done this. It could be improved, I'd love to see it running over ssh tunnels to a file server, for example, but hopefully that will appear later, rather than requiring an attached hard drive. I guess it can backup to mounted shares though?

Re:I'm too lazy to do any research...

[snark23]

Yeah, you're right, it doesn't use ZFS. I got that idea from an ArsTechnica article, which I didn't read all the way through. The article was actually arguing that Time Machine should use ZFS :-P

Re:I'm too lazy to do any research...

[paanta]

No, it cannot gracefully backup to mounted shares (unless they're on another 10.5 machine). Seems like something that apple or a 3rd party will address soon, though...

Time machine really is slick. I hook up to a usb hub and big LCD when I'm at home with my macbook, and it uses that opportunity to back things up to the usb drive I've got sitting on my desk, so there's a nice, seamless backup created every evening at the very least. Whine about how time machine is nothing new, nothing fancy, etc as much as you like. I'd be interested to know how many slashdot types back up their computers every night. I'm guessing the answer is 'not many'. And these are people that know better.

Re:I'm too lazy to do any research...

[shmlco]

ZFS-based block-level backups would really only help in one specific type of situation: A large database where specific blocks (and only those blocks) change. Otherwise, if I have a very large Word document and do something as simple as add a word to the beginning of the file, then every block's contents have shifted and I still need to backup the whole thing anyway.

Then again, if you're putting huge databases on your drive then Time Machine probably isn't your solution anyway (nor is rsysnc).

Re:I'm too lazy to do any research...

[Anonymous Coward]

You must have a sad, empty life if you have such a narrow definition of "sexy". I'm very sorry for you.

Re:I'm too lazy to do any research...

[nschubach]

Didn't you know? Sex is the only acceptable hobby. Men must sell their souls for it, women must regulate it and be rewarded before doing it with shiny stones and food, and anyone that defies that process is a sad and/or pathetic person. You must promote this thought in adolescents. They must be taught these rules or be hung. Early teen abortion is acceptable for anyone who does not comply. You may as well commit suicide because your life is not complete without it. Especially if your bald, fat, hairy, etc. You should just give up now because you will never be "cool" and people will hate you for not having it. They can tell by looking at you if you've had it. If you don't have it at least every other day, you will be lynched like the evil monster you are. Shame on you!

Oh, sorry, I had a social acceptance moment there.

Ghost

[Steauengeglase]

Couldn't I just stick that Ghost for Linux CD I burned last year in before I go to bed and get the same results?

Re:Ghost

[Constantine XVI]

Ghost isin't automated, it can't restore single files, it's not exactly easy to use, etc.

Time Machine is for recovering old versions of files on a whim.
Ghost is for restoring an entire system to a certain state.

So ...

[kalidasa]

Like TimeMachine, can this restore multiple versions of the same file? Did you use ZFS? Or is this just a GUI front end for a simple rsync backup?

Re:So ...

[CapitanMutanda]

Like TimeMachine, can this restore multiple versions of the same file? Did you use ZFS? Or is this just a GUI front end for a simple rsync backup? VMS was doing versioning ages ago... but the GUI does the trick apparently

Re:So ...

[Lachryma]

Time Machine does not yet use ZFS.

Re:So ...

[dsginter]

Think of it this way:

Rsync is to data what duct tape is to... well, everything else: it might not be pretty on a visual basis, but you'll be damned to find a better solution on a bang/buck basis.

Most geeks are pretty happy with duct tape and rsync. This will be difficult to change because geeks, nearly by definition, can see beauty beneath an ugly fascia.

Re:So ...

[Anonymous Coward]

No - geeks are stuck with an uglier fascia, and convince themselves that it is somehow 'better' to avoid a lifetime of anguish and jealousy.

Re:So ...

[kalidasa]

I know time machine does not yet use ZFS, but snapshotting would get you something like TimeMachine with an open file system.

Re:So ...

[gardyloo]

This will be difficult to change because geeks, nearly by definition, can see beauty beneath an ugly fascia.

    Fascia = connective tissue (usually) or an architectural term. I prefer (in this context) the "belt" definition :)

    Façade = face.

Re:So ...

[neuro.slug]

...geeks, nearly by definition, can see beauty beneath an ugly fascia.

That explains their girlfriends.

*ducks*

Makes you wonder ...

[ThirdPrize]

what apple users did for backups before version 10.5 of their operating system? I just drag my important files onto an external drive.

Re:Makes you wonder ...

[tulmad]

Most of us just dragged our important files to an external drive

Re:Makes you wonder ...

[Just Some Guy]

Most of us just dragged our important files to an external drive

Yep. The biggest advantages that I see are that you don't have to remember to do it, and that there's one central, (presumably) well-maintained backup solution instead of a million home-rolled automation attempts.

Re:Makes you wonder ...

[drerwk]

I've used Retrospect since 1990. Terrific product. So good I haven't had to buy an update for years. Probably why Dantz was bought.

Re:Makes you wonder ...

[dbzero]

Before Time Machine, I made two bootable backups. One to an external firewire drive and the second to an internal drive. When the software I used--SuperDuper!--updates to Leopard, I'll go back to doing that to at least one drive and possibly continue using Time Machine for the other.

Re:Makes you wonder ...

[slyn]

SJobs claimed in the keynote when he unveiled Time Machine that 80% knew they should back up regularly, 26% did backup data, but only 4% did backups on a regular basis.

I got those numbers from here, but those numbers were first released (made up? ;P) at WWDC earlier this year IIRC.

Re:Makes you wonder ...

[Bert64]

It it possible to get time machine to back up over the network, to a non apple server...
I like the idea of automatic backups, but all my macs are laptops so having an external disk permanently attached is a nuisance. I would like backups to be performed whenever i'm attached to my home network, and stored on a linux server that has a stack of large drives.

Re:Makes you wonder ...

[y86]

You still need your external drive.

• If your internal drive burns a bearing you don't lose your data you just lost 12 versions of it.
  
• This not an external backup tool, it's a versioning tool, like easy CVS for your data.
  
• It's an awesome idea but it does not replace external backups.
  

Re:Makes you wonder ...

[jellomizer]

No it is an external backup tool with versioning tools like functionality. If you use Time Machine as a Versioning Tool you will probably get burned. Because chances are there is equal chance you backup drive will fail before your primary drive... Chances are Extremely slim that both die at the same time. But if your External Drive dies and you get a new one you will backup you system to the state it is currently....

Re:Makes you wonder ...

[99BottlesOfBeerInMyF]

Makes you wonder what apple users did for backups before version 10.5 of their operating system? I just drag my important files onto an external drive.

Backup.app worked just fine for incremental backups to DVD, network drives, etc.

Re:Makes you wonder ...

[ElGanzoLoco]

I've been using the free Carbon Copy Cloner for a while to do bootable backups of my entire hard drive. Crude, but it works well and I currently don't really need other features. The only downside is that it can take forever to back up everything.

When I'm away from my backup harddrive, I do just as you say and drag and drop my most important files to my iPod.

Of course, once I upgrade to 10.5 I'll probably wonder how I've been living without Time Machine all this time. Plus ça change...

Re:Makes you wonder ...

[GaryPatterson]

SuperDuper and an external Linux-based NAS running AFP.

Re:Makes you wonder ...

[pebs]

It it possible to get time machine to back up over the network, to a non apple server...

Is that a question? Because I'm wondering the same thing.

There is Linux support for HFS+ and AFP which may help achieve this, but I'd imagine they would need to update HFS+ support with the new filesystem changes and probably update AFP support as well. Maybe its possible to do some hackery and get it to work with an updated AFP and a typical Linux filesystem like ext2/3, considering that AFP works fine with Linux filesystems right now, and hard links are already supported. It would just be a matter of making the changes to AFP to support hard links.

Just my take on it, there may be some details that I'm missing.

How is this different than Shadow Copy?

[tunafreedolphin]

How is this different than Shadow Copy?

Re:How is this different than Shadow Copy?

[Anonymous Coward]

Shadow Copy only allows you to restore a whole drive as opposed to just single files.

Re:How is this different than Shadow Copy?

[Anonymous Coward]

that isn't the case. shadow copy allows you to view any previous state of a folder that's in the shadow copy db and restore any single file. when browsing a shadow enabled folder from XP or Vista, you can choose to 'view' older versions of a folder rather than 'restore' and from there you can drag and drop any files you'd like.

Re:How is this different than Shadow Copy?

[dave420]

It's a different implementation of pretty much the same thing, bar shadow copy's shadow volume use, network awareness, restore functionality, etc.

Re:How is this different than Shadow Copy?

[Anonymous Coward]

Why don't you read up on it and find out you lazy bitch?

Re:How is this different than Shadow Copy?

[EvanED]

One important difference that I've only recently picked up is that Time Machine is actually intended for backup purposes: you have to do it on a separate drive. Shadow copy, I believe, works on the same volume.

What you're using it for will greatly influence which of those approaches is right for you. If you want a backup, Time Machine is more what you want, because it protects you against drive failures. If you're like me and want a versioning file system, shadow copy is much closer to what you want, though still not completely there.

(I was all excited that features like what I want are finally becoming somewhat commonplace, being in both Vista and OS X 10.5, but was disappointed to learn that Time Machine isn't really what I want at all. Oh well...)

Completely misses the point

[Just Some Guy]

We've had backup systems for decades. Even Windows has a more functional system than Leopard by accounts I've read. What Leopard did is make backup and restore sexy to the point that people will actually want to do it.

"Flyback" is a replacement for, well, I'm not sure what. It's certainly nowhere near Time Machine whose primary innovation was "damn gotta get me that" user-friendliness.

Re:Completely misses the point

[santiago]

What Leopard also did was add file system support for hard links to directories, so that backups from different points in time could be easily presented as complete volume images without any need for a special backup file format, yet still share storage for unchanged files.

Re:Completely misses the point

[Sosarian]

RSync also has this, tools like dirvish take advantage of it.

Re:Completely misses the point

[Anonymous Coward]

Forget about sexy, Leopard makes it easy. I used to have a haphazard system involving rsync running off a script copying to an external drive. Now I bought an internal drive and let the OS do everything for me. Could I have hacked together some system to work off an internal drive before? Sure. Did I? Hell no. And that is what makes all the difference.

Re:Completely misses the point

[Just Some Guy]

I used to have a haphazard system involving rsync running off a script copying to an external drive.

I hear you, and said that in another post: having one standardized app for something so fundamental is much better than supporting a million half-assed ones.

Re:Completely misses the point

[deander2]

i do plan on adding the "damn gotta get me that" user-friendliness. i just wanted the get the basics done first. it works the same basic way already, where you can move "back" (ie, to previous snapshots) in the same directory. i just have to learn the python-opengl bindings.

Re:Completely misses the point

[empaler]

It's a FOSS implementation of Time Machine. And it's v.0.2. They're probably going to get to the slick "gotta-get-me-that"-UI within a few years. This is just building off of rsync, so it's just a nicer UI for the users, not something new and revolutionary.

Re:Completely misses the point

[Just Some Guy]

And it's v.0.2. They're probably going to get to the slick "gotta-get-me-that"-UI within a few years.

I understand all that, but why then is it on the front page of Slashdot today?

Re:Completely misses the point

[hoggoth]

> i do plan on adding the "damn gotta get me that" user-friendliness.

Well, if you are serious about adding functionality like Time Machine, please realize that doesn't mean a prettier interface. That means you are adding a window menu option to every application window to enable browsing the backup archives. When the user chooses that option, you must trap every file system interaction from the processes in that application and serve up data from the archive instead of from the "live" disk. You must do this to an application right in the middle of it's execution substituting it's already open file handles for your own and preserving the state of every open file somehow even though they are now connected to different files...

It boggles my mind how Apple did this...

Unless Time Machine only works with certain modified applications, and not any and every application?

Re:Completely misses the point

[deander2]

it only works with certain apps, from what i understand. for a detailed writeup about it, read the article on arstechnica. it's pretty simple.

i do plan nautilus integration....

Re:Completely misses the point

[Anonymous Coward]

Sorry, but nothing to do with computers or electronics or politics is "sexy".

Even Apple products.

(OK, you could be pedantic and start going on about porn but that would just be twisting the language to argue a point).

Some people need to get out more and see real people, then they will see what I mean.

Re:Completely misses the point

[beh]

speaking of the sexy interface - it's one of the things more likely to piss me off about it - the animation of a wormhole (presumably) and the moving stars are just plain annoying... ...not quite on the scale of the paper-clip, but annoying nonetheless...

(oh - and unlike the paperclip, this one is limited to one particular screen and doesn't just 'jump up' because you use a function that isn't used every other day)...

On the positive side - what Apple HAS achieved with it, is that people begin to take backups more seriously, and even so far (even if just for playing around with it) that in the beginning they might create a few files and then occasionally modify them and finally delete them, just to see how easy it is to get the file back... (personally, for MOST backup software I have used, I have not really gone through the trouble of trying to seriously create, backup and restore some files 'just for the fun of it', because getting the data back is almost as painful as creating it in the first place. (and I simply trusted the backup solution to actually work -- the one time I actually needed an amanda backup restored a few years back, at first I had to spend 20 odd minutes looking through the documentation on how to actually do it (and I did exactly know which file I needed). The one time I tried it with TimeMachine, it just DID THE RIGHT THING (tm)...

Re:Completely misses the point

[hoggoth]

Yes, I see now. It works with Finder and some key applications - which is great. But I thought it would work with ANY application, even ones that weren't specifically "time machine" aware.

So if you wrote a nice Nautilus integration and an API for other apps to access "Time Warped" files that would be equivelant to what Apple has done (within the limits that you don't have control over the other apps)

Re:Completely misses the point

[dbIII]

Even Windows has a more functional system than Leopard by accounts I've read

No. Marketing hype and reality differ - NTbackup has been around for years but it usually gets replaced after the first recovery effort with one of dozens of products that show more signs of being able to do the task. Even now complete disk imaging from what is effectivly a different operating system (eg. acronis, ghost, etc) is the only truly reliable way to be sure you get it all. Everything bar the users mailbox and registry is just not good enough and there are so many edge cases that block access to these or give you partial corrupted copies.

Re:Completely misses the point

[Just Some Guy]

I was referring to the Volume Shadow thingy. I don't use Windows but I've heard really great things about it, except that it's very poorly marketed and not really integrated with anything.

Re:Completely misses the point

[sciurus0]

True, but since on linux you can't create a hard link to a directory even on a snapshot with no changes dirvish has to recreate the entire directory structure then create the hard links to files. This is slower and wastes space compared to Apple's approach.

No Open Source Invovation here!

[jellomizer]

I hate to sound like a fanboy but...
Apple did spend many months working on the interface and desing to try to make backups as
easy as possible... All the time they took was really in design time... A much smaller portion
was used in actually coding.... (Find files that have been altered from last update -> Copy Said files
to alternate drive in directory with the date as a name, make note of files that have deleted)

To Restore data go to the date of backup when data existed merge with previous dates and account for
deleted files.

Once selected copy files back to origional drive...

It really isn't a complex process... And I am not supprised that someone made it for Linux
within a couple of weeks of Leopard being public...

Apple did all the design work which was actually the hardest part the programming isn't that hard.
I would be careful for patent issues though... Apple is a big pattenter... (Espctially after
Microsoft stole their interface)

Re:No Open Source Invovation here!

[ickoonite]

(Find files that have been altered from last update -> Copy Said files to alternate drive in directory with the date as a name, make note of files that have deleted)

Trivialising the technical underpinnings of Time Machine is unwise, and plays right into the hands of those who say Apple is all about show and lacks substance. In fact, the way Time Machine knows what files have been modified is really quite elegant and shouldn't be underplayed. I shan't go into the details of it all here, but if you are interested, see the relevant page of John Siracusa's excellent review of 10.5 over at Ars Technica.

In the meantime, you might like to consider learning how to spell.

:|

Re:No Open Source Invovation here!

[thegnu]

Apple is a big pattenter... (Espctially after Microsoft stole their interface)

Which they stole from Xerox. Funny. I saw Woz speak, and he stated the following:
a) they toured Xerox, and saw everything they did, went home, and made it for cheaper
b) Windows "stole" their interface
c) The Creative Labs suit about the iPod interface was silly and unfounded

Hmmm? So any lawsuits AGAINST Apple are silly and unfounded. Those same lawsuits file BY Apple are great and wonderful, huh? Can someone explain this to me?

By the way, I think patenting obvious ergonomic ideas is stupid, and it's not theft to copy an interface. The same way making a car with a steering wheel on the left side is not theft of intellectual property. Apple has, IMO, more mud on their faces because they have the balls to turn around and call themselves moral. And as our president says, "You can't claim the high horse and take the low road."

Re:No Open Source Invovation here!

[jellomizer]

Trivializing good software design is unwise. As programmers we go ga-ga over something that we think wow I doubt I can code this easilly, the person who made this musta been really smart. The design process and getting even the simple stuff to work well is actually a lot of work more then you think. Normally as programmers we copy what people do, heck lately for my photoshop work I have been adding reflections because looking at the picture of the reflections I see how to do it in photoshop Copy Layer, Flip Virtically, 35% transparancy. New Layer Gradiant fill from Background color to transparent over the object. Easy to do, but I never done it before because I never though of trying to do it... The same with back in when I did ANSI Art for BBS's After seeing the Wolfinstine 3D graphics for the logo I figured out how to make my ANSI look like Metal. Before having experience with GUI I envisioned computers of the future (Um now today) to look completely different. Except for windows I was thinking more of frames. The design part is where the real inovation is, once someone has a good design programing it, is very simple... The part that makes jobs are programmers hard is the fact there is rarely a design team so they need to do the design work as well. There are actually very few things that will take massive programming skills to complete, things I would give programmers a lot of credit are things like Googles search algorithm, Graphic Tools healing brush... But most things are just because of good design, and perhaps coding with a lot of detail but with skills that anyone with a CS Degree can accomplish.

Re:No Open Source Invovation here!

[jdgeorge]

Informative, but condescending to the point of insult. However, I have no mod points, so I can't mod either way. Instead, I will make a few comments:

Trivialising the technical underpinnings of Time Machine is unwise...

Hmmm... unwise? Perhaps uninformed, but surely not unwise, unless...

...and plays right into the hands of those who say Apple is all about show and lacks substance.

Plays right into their hands, eh? Well, then those people would have had their way with him, wouldn't they, had not some heroic Slashdotter stepped in to prevent that disaster by informing the poor ignoramus. It appears that we can consider this crisis averted.

The links are helpful and informative, thank you.

Re:No Open Source Invovation here!

[jellomizer]

No I wasn't supporting Apples Patents I was just stating after Microsoft took their design. Apple just started getting more serious about protecting their patents, to prevent that action happening again. Yes we all know apple got the original Idea from Xerox, but Apple was actually using the idea and made it so it was useful and affordable, value add. Then Microsoft took and used the interface without adding value to it...

I beleave Apple lost the Creative Labs suit and pay them royalties to use the patents. Wich they should if they stole the patents. There is a large issue of patents then who are the Patent good guys and bad guys, The Patent system for software is broken allowing to many patents for simple things.

Re:No Open Source Invovation here!

[DragonWriter]

Trivialising the technical underpinnings of Time Machine is unwise, and plays right into the hands of those who say Apple is all about show and lacks substance.

Emphasizing that the main work of a project is in developing the UI is not the same as trivializing its technical underpinnings (the two may or may not go together), and, even when it does involve trivializing the technological underpinnings, isn't the same as saying that it lacks substance. What people with a technical focus (hi, Slashdot) sometimes forget is that the substance of a product isn't just the theoretical capabilities of the engine, but what it delivers to the user. Ease of use and intuitive interface that make the underlying engine usable to non-specialists are not "show", they are substance of a kind that is just as important as good technical underpinnings.

Re:No Open Source Invovation here!

[Anonymous Coward]

Apple is a big pattenter... (Espctially after Microsoft stole their interface) Which they stole from Xerox. Funny. Whoops, you're inventing history.

Xerox licensed the rights to Apple in a stock-swap deal in 1981. That's why Xerox never pursued after Apple. However, Xerox was very interested in the Microsoft case... as a Microsoft violation of Apple's licensed intellectual property could have implied a Microsoft violation of Xerox's intellectual property.

Re:No Open Source Invovation here!

[mrbluze]

I hate to sound like a fanboy but...

I'd hate to sound like someone who is praising someone who sounds like a fanboy but...

Most of this thread reads like a bunch of people stumbling and saying either:

• It's just a cheap gimmick with a pretty face
• Don't worry, we'll have something like this for Linux in no time
• Mine already does it and I didn't have to pay for it
• Oh.. they did something clever? Well I for one think doing something clever is stupid, so there! {promptly puts head back in the ground}

The plain fact is you're right - Time Machine is a good product and is a credit to Apple. My only dilemma is whether to fork out the money and install Leopard, or to do what I was going to - discard my Tiger installation and install Linux as the sole OS on my macbook. As much as Apple has bells and whistles, I don't trust them any more than Microsoft. It's just that Microsoft is not as good at hiding their built-in backdoors.

Re:No Open Source Invovation here!

[99BottlesOfBeerInMyF]

I would be careful for patent issues though... Apple is a big pattenter[sic]... (Espctially[sic] after Microsoft stole their interface)

I always wonder about Apple's reputation as a danger to innovation via patents. They seem to patent a medium number of technologies, usually interface elements and as near as I can tell almost never bring suit against others over patents, although they do occasionally for copyright or trademark concerns. From what is reported in the news, however, it seems that people are suing Apple, not being sued by them.

Re:No Open Source Invovation here!

[hmccabe]

In the meantime, you might like to consider learning how to spell.

Nevur!

Re:No Open Source Invovation here!

[xirtam_work]

Xerox got Apple stock in return for licencing their intellectual property, which Xerox saw little commerical value in and had been sitting on for years.
Microsoft stole Apple's implementation of 'WIMP' after becoming an Macintosh developer and snaffling all the documentation. It still took them over ten years to produce what Apple did in two.

Re:No Open Source Invovation here!

[Anonymous Coward]

Wow, you're the biggest asshole ever.

hard link directories

[Ydna]

To make it really work like Leopard's Time Machine, we need a way to create hard linked directories. I mean besides the obvious ones that are made for us. Otherwise you get massive trees of directories containing hard linked files (for those that have not changed).

It's easier to just use rsnapshot.

Re:hard link directories

[sammy baby]

Heh. Just be careful how you go implementing that, or you could wind up with problems like these.

Re:hard link directories

[remmelt]

If I remember correctly, rsnapshot saves diffs of files. This is good for space reasons and works well with text, but not so with large binary objects, which will probably be stored as a new copy. This negates the space profit. Macs tend to be used by people involved with video/graphics/music.

Re:hard link directories

[Ydna]

rsnapshot uses hardlinks between files that are unchanged between backups and full copies of changed files. Each "snapshot" will look like a fully formed copy of the original. It's all done using rsync.

Ubuntu TimeVault

[phoebe]

There's already been work on a Linux Time Machine, just not ready for prime time yet: TimeVault.

Re:Ubuntu TimeVault

[ceeam]

BTW, except of Mark's past, why do they serve almost everything over HTTPS?

Re:Ubuntu TimeVault

[DragonWriter]

BTW, except of Mark's past, why do they serve almost everything over HTTPS?

Just curious, is there any really good reason not to serve everything over HTTPS?

Re:Ubuntu TimeVault

[CarpetShark]

Actually, this is already finished. Once it hit milestone version 0.34, it went to the future, to bring back it's 3.0 version.

Re:Ubuntu TimeVault

[growse]

Some would argue that latency is a good reason. Unless you can offload the SSL using some sort of special hardware, and latency is *really* important to you, you wouldn't use HTTPS.

Re:Ubuntu TimeVault

[Kjella]

And finally a good reason why Linux should have a "My documents"-like folder under home:

Watching /home

I'm trying to watch my home directory and deleting a file on my desktop results in a dozen changes.

Right. Don't do that.

Watching the /home directory may result in hundreds of unnecessary snapshots a day because many programs use temporary files that are created under you home directory. The good news is that, usually, the file is gone by the time TimeVault gets around to snapping it, and so an entry is made that a file changed, but no snapshot is made.

Re:Ubuntu TimeVault

[Anonymous Coward]

And finally a good reason why Linux should have a "My documents"-like folder under home: mkdir ~/"My Documents"
Done!

If you want a My Documents folder, make one.

Re:Ubuntu TimeVault

[Anonymous Coward]

mkdir ~/My\ Documents

There. That was hard.

Re:Ubuntu TimeVault

[Blakey Rat]

It adds anywhere from 0.5-1.5 seconds of latency to a HTTP connection. If you're serving a webpage with gobs of small files, you'll increase your load time quite a bit by serving it as HTTPS instead of HTTP. Of course you could take a mixed approach, but then you get a warning in most browsers that the page contains mixed HTTP/HTTPS data.

It would be hard to build a case for serving a site like, say, Wikipedia or Archive.org or even Google (search, not their other apps) over HTTPS.

Cool

[Rik Sweeney]

Think I might install this tonight. At the moment I just have a crappy little script that copies my entire home directory onto a backup drive, overwriting anything that has changed.

Looks like the Hungry Hippo release of Ubuntu will have a new application soon.

Re:Cool

[ThrobbingGristle]

As mentioned elsewhere, I would look at rsnapshot or rdiff-backup. Scripts (no GUI, that I know of) but both wonderful in their own way.

Re:Cool

[arose]

Pybackpack would be a GUI for rdiff-backup.

Not the interface

[Thornburg]

IMO, it is not the _interface_ that is cool about Time Machine, but the ease of use and the fact that it is fully automatic.

I didn't RTFA, so I don't know if this "Time Machine for Linux" implementation is as easy to use or not, but the real thing that makes Time Machine cool is that even my mother can use it.

The Ars Technica article about Leopard has lots of very cool details about Time Machine in it, including how it works. (It uses hard-links, including hard-links to directories, so in each and every time-stamped folder on the backup drive, you have a *FULL* copy of your HDD at that time (minus anything you excluded from the backups). Read that portion of the Ars Technica article if you want answers to questions about it.

Re:Not the interface

[Hatta]

IMO, it is not the _interface_ that is cool about Time Machine, but the ease of use and the fact that it is fully automatic.

What's the difference? The interface is how you use software. If it's easy to use, it has a good interface.

Re:Not the interface

[MouseR]

Yeah. There's more to Time Machine than just a one-off backup of your data. TM aggregates changes and you can roll back to any point in time.

Think of it as CVS. It tages backup times but actually only copies new data as it's checked in.

Also, TM is not confined to the Finder per say. if you're in Address Book and lost a contact, type in the filter string to locate it. Still can't find it? Right there from Address Book, hit Time Machine and Address Book will be served with backed-up address book data, filtering on the fly, as you go back in time until you find what you've been looking for.

Same thing for anything spotlight-able.

So, yeah, it's got a pretty interface, but TM goes way beyon just file/backup management.

Re:Not the interface

[slashflood]

It uses hard-links, including hard-links to directories, so in each and every time-stamped folder on the backup drive, you have a *FULL* copy of your HDD at that time (minus anything you excluded from the backups

This is exactly how BackupPC works! The interface isn't as fancy as Time Machine (because it's web based), but even the workflow is the same. It is fully automated and you don't have to touch anything. As soon as your notebook is connected to the BackupPC server, it starts to make an incremental backup. The restore is as simple as selecting the date, the directory and clicking on a button.

Re:Not the interface

[robot_love]

IMO, it is not the _interface_ that is cool about Time Machine, but the ease of use and the fact that it is fully automatic.

I didn't RTFA, so I don't know if this "Time Machine for Linux" implementation is as easy to use or not, but the real thing that makes Time Machine cool is that even my mother can use it.

So it is the interface, then?

I realize the interface doesn't do the heavy lifting in an application, but I wish the FLOSS crowd would finally clue in to the fact that ease-of-use matters. For example, GnuPG is a way to protect your privacy through encryption, but it only has a CLI. GUIs exist for GnuPG, but their installation is complex. Why do people work on GnuPG? Because privacy is important! But who gives a fuck when only 1% of the population can use it? Thanks for nothing, GnuPG!

I have no particular bee in my bonnet about GnuPG, it was just the latest FLOSS effort to piss me off. Open-Source software and "Free as in Freedom" are ideas too important to be relegated to the technical elite, but the technical elite's refusal to make their tools easy enough for the rest of us cuts out most of society. You have the cure for cancer but refuse to give it to us because we don't have the time or desire to learn Perl.

This Linux "Time Machine" sounds cool. Too bad I'll never be able to use it. Bah!

Re:Not the interface

[Anonymous Coward]

Too long have I sat by and watched a great phrase be abused:

...Also, TM is not confined to the Finder per say. if you're in Address Book and lost a contact, type in the filter string to locate it. Still... It's PER SE, goddamnit! And it means "intrinsically!" You saying "TM is not confined to the Finder per se" would imply that either it IS somehow confined to finder (but not intrinsically) or you just like to use big-person words you don't understand. Please! Think about what you're saying!

Re:Not the interface

[Anonymous Coward]

Please! Think about what you're saying!

OK, I've thought about it: You're an asshole.

Thanks! I feel much better.

Re:Not the interface

[MightyYar]

I think that you are being too hard on the open source crowd. There are basically two main contributors to open source:

• Companies
• Geeks

Now, the companies are usually trying to sell service contracts to IT departments, so grandma is not in the game plan. Linspire pops to mind as a notable exception. Another is Limewire. Both of those companies arguably sell a pretty usable product for the non-IT person.

The other group is geeks. Some high percentage of the time, an open source project exists because some dude made a tool for himself and then figured that other people might like it. By definition, these tools are going to be directed toward geeks - not laymen. Only when it gets SO bad that even the geeks get pissed off does something actually get done, and so you get projects like Ubuntu and Gnome.

For the most part, these people are not taking your money, and it's not really that important to them whether or not you decide to use their software. I hate to sound like a Linux zealot, but if you don't like the free tools offered you can either walk, fix it yourself, or pay someone to fix it. You can bitch at the developers, too, but they probably won't care much - and who can blame them? The tool obviously fits their needs.

Personally, I like Macs because the computer is very polished and usable straight out of the box, but you can drop to a command line and X11 and play with most of the open source world's toys. You can even boot into Linux if you like.

Re:Not the interface

[bestinshow]

That's great if you know the date you want to restore too.

Time Machine's ability to simply browse backwards through time in the folder, whilst still having the folder functionality usable is far beyond BackupPC. Indeed I bet there are many times that you just want to do this, you don't want to restore the file or the folder as it was then, you just want to quickly glance inside the file as it was.

There's nothing amazingly clever about Time Machine, but it is Apple "Getting It Right(tm)" interface-wise (excluding silly starfield, etc) and functionality-wise.

Re:Not the interface

[the phantom]

And yet, the FOSS community seems to think that everyone should be using FOSS. You can't have it both ways -- either FOSS is only for geeks and large corporations, in which case it will never catch on with the consumer; or it is for everyone, in which case the geeks advocating FOSS need to make sure it can work for everyone. Otherwise, comments about how everyone ought to be using FOSS are hypocritical.

Re:Not the interface

[Anonymous Coward]

It's called Seahorse, and you can install it from the mainstream repositories.
apt-get isn't very complex installation in my book.

Also try this: http://www.gnupg.org/(en)/related_software/frontends.html

Re:Not the interface

[mrv20]

> You have the cure for cancer but refuse to give it to us because we don't have the time or desire to learn Perl

I think you've got that backwards. Exposure to Perl almost certainly causes ill health, not cures it.

Re:Not the interface

[Anonymous Coward]

Do mac fans know of any other adjectives besides 'cool'?!
You sound like a bunch of 8 year olds!

Oh, and I can't read the fkn captcha!

Now get off my lawn!!!

Re:Not the interface

[Anonymous Coward]

Is that spinach in your teeth?

Yeah. It's latin, but a formula only used in english as far as I know. So I misspelt it, fooled by numerous occurrences of it here and there. Sue me.

When you write something in french half as good as I do in english, I'll accept the nagging tone (and your utter lack of balls for not signing). Meanwhile, go find yourself a grammar web site and bitch all you want. This isn't the place for it.

--mouser

Re:Not the interface

[goofy183]

This is called rsync with hard links. I have a shell script I've been using for a few years that does nightly backups of my machine to another via SSH (even works when I'm on the road). The script uses hard links to create a minimal backup each time while saving disk space and making recovery easy. I can go into any one of the timestamped directories and see a 'full' view of all my files as they were on that day.

To see said script: http://erics-notes.blogspot.com/2007/10/poor-mans-timemachine-rsync.html

Re:Not the interface

[Trailer Trash]

Please! Think about what you're saying!

Isn't that supposed to be "seing"?

[duck/run]

Re:Not the interface

[BobPaul]

The interface is the goofy 3d zooming through space view. The ease of use and he's referring to is that the incremental backups that are stored all appear like full backups from a file system perspective.

One can set it up once with the goofy 3d zooming thingy and then it'll happen automatically in the background. Need a file that you know was good a week ago? In term, type "cp /mnt/backup/11-01-07/path/to/file/file ~/" etc. Or browse to it in the finder. Or use the goofy 3d zooming thingy.

Most backup solutions require you use their software to restore from backup. If I interpreted the parent right, Leopard doesn't.

Re:Not the interface

[Anonymous Coward]

And yet, the FOSS community seems to think that everyone should be using FOSS. You can't have it both ways -- either FOSS is only for geeks and large corporations, in which case it will never catch on with the consumer; or it is for everyone, in which case the geeks advocating FOSS need to make sure it can work for everyone. Yeah, the problem is that ones writing those programs are usually different than those trying to get everyone to use FOSS...

Re:Not the interface

[vadim_t]

Huh? There's integration of GPG about everywhere.

KMail has very good support for it. Thunderbird with Enigmail works nicely too.

The problem with crypto is that it's hard to understand. You can create a key, set up kmail, and teach your grandma to enter the passphrase when asked, but that won't be secure. Public key crypto needs the users to follow the proper procedure for exchanging keys and verifying signatures, and explanations of how all that works sounds terribly boring to many people.

Re:Not the interface

[andre.ramaciotti]

We use 'per se' in Portuguese too, though we write it as 'per si'.

Re:Not the interface

[Anonymous Coward]

Oh good God. Thank you very much! Thank you! I'll return the favour when people use 'effect' instead of 'affect'. I promise.

Re:Not the interface

[Anonymous Coward]

I wish the FLOSS crowd would finally clue in to the fact that ease-of-use matters

well then you'd better fire up your time machine and use it to travel forwards from the year 1999 to join us in 2007 when the "FLOSS crowd" has in fact "cottoned on" to that fact for several years. But perhaps you're talking about a different "FLOSS crowd" to me, because like you, i was also referring an arbitrary subset of all known FLOSS software developers and then generalising their attributes to the rest of the set for no good reason, with the only difference being that I chose to selectively focus on the ones who get it right, whereas you have selectively focused on the ones that get it wrong.

hmm... wait a minute.. i just realised something! arbitrary generalisations are FUCKING STUPID.

Re:Not the interface

[tietokone-olmi]

The thing about GnuPG is that you generally want to use it in conjunction with an E-mail application. Thus rather than fancy GUIs for GnuPG, we have interface libraries so that e-mail program authors can facilitate integration into their programs.

See libgpgme (GPG made easy) for instance. It's quite simple for something as complex as public key encryption.

Still, I'd quite like to have a snazzier key management program, since it's a bit overkill to have every MUA reimplement that.

Re:Not the interface

[General Lee's Peking]

Maybe I just need this explained to me because I'm having difficulty conceptualizing this: how do you separate the idea of good interface design from the idea of ease of use? Maybe you might have some example to help me see this because I can't think of one piece of software that's easy to use whose facility didn't have everything to do with the interface.

Re:Not the interface

[jav1231]

It's not just a goofy-3d-zooming-thingy. You can zoom through space, find the image you want, then drag-n-drop items right through the goofy-3d-zooming-thingy. That makes it a cool-as-shit-goofy-3d-zooming-thingy that's also functional. BTW: tell your Mom to type "cp /mnt/backup/11-01-07/path/to/file/file ~/" and she'll probably wash your mouth out with soap! Tell her to find what she deleted and drag it to her desktop and Mommy will bake you brownies! Mmmmm...brownies!

Re:Not the interface

[Anonymous Coward]

Half as well.

And when I use incorrect French on a French website, you're welcome to correct me. I'd rather write correctly, thanks.

Re:Not the interface

[Anonymous Coward]

Which reminds me... I know I had your mom's phone number here someplace.

*ZOOOM THROUGH TIME*

That's right. See you at 8, kiddo.

Re:Not the interface

[toph42]

Are you that uptight per se, or just on Slashdot?

Re:Not the interface

[Burz]

The Ars Technica article about Leopard has lots of very cool details about Time Machine in it, including how it works. (It uses hard-links, including hard-links to directories, so in each and every time-stamped folder on the backup drive, you have a *FULL* copy of your HDD at that time (minus anything you excluded from the backups). Read that portion of the Ars Technica article if you want answers to questions about it. To elaborate, what is normally "kinda efficient" using rsync and rsnapshot, is made really quick and efficient in Time Machine because snapshots don't have to be compared and linked at the single-file level.

Re:Not the interface

[Anonymous Coward]

> You saying "TM is not confined to the Finder per se" would imply that either it IS somehow confined to finder

It's YOUR saying, goddamnit!

Re:Not the interface

[Anonymous Coward]

Just keep waiting for it then... Ah! here's one now!

Re:Not the interface

[Mr2cents]

I'm using rdiff-backup backup myself, it seems that it does the same as your script. Every night an incremental backup is made, and when the backup server is done, the desktops are turned off (and the server itself).

Re:Not the interface

[gravis777]

So in other words, its a backup with system restore points with a pretty interface.

You know, I am using Leopard, and just fired up Time Machine. I do not see a way to create a backup unless you have a secondary disk. It looks like you point it to a backup drive, and it creates a disc image of your computer at regular intervals.

Hmmmm, this reminds me a LOT of Deja Vu, which I have been using for a while to create daily backups of our XServe to another location (in addition to the RAID that is built into the server, which I do not entirely trust as its a software raid). So, can someone tell me, featurewise, what is the difference?

I know my co-worker was talking about all these new features in Leopard, and I was like, um, Windows (or third party windows apps) has been doing that for years.

Re:Not the interface

[SallyShears]

...then drag-n-drop items right through the goofy-3d-zooming-thingy... Not quite... You find the the file you want on one of the backups, then click a "Restore" button. You cannot drag and drop from the backups.

All in all, it is nifty, automatic, and relatively painless. Still waiting to discover the quirks... Bound to be a few in something this new and complex.

Re:Not the interface

[MouseR]

"Writing", actually ;-)

Re:Not the interface

[Frantactical Fruke]

Things we did not want to know #42: The reason Finnish speakers NEVER have problem spelling "per se", even if English is their fifth language, is that in Finnish "perse" is a rude word for "posterior".
A similar reason explains why people named Pascal or Pascale never stay long in Finland...

Re:Not the interface

[TheNetAvenger]

Not quite... You find the the file you want on one of the backups, then click a "Restore" button. You cannot drag and drop from the backups.


Sadly you can drag and drop from Vista's 'Previous Versions' (snapshot/backup) interface...

Again I submit, Vista not only does more than Time Machine, but is easier to use.

Re:Not the interface

[Anonymous Coward]

People are actually using vista/longhorn/2003?

Re:Not the interface

[Anonymous Coward]

It's God damn it, goddammit!

Re:Not the interface

[Thrip]

ease-of-use matters. For example, GnuPG is a way to protect your privacy through encryption, but it only has a CLI. A lot of us consider a CLI easier to use.

the technical elite .... have the cure for cancer but refuse to give it to us because we don't have the time or desire to learn Perl. Perl? Perl? That's your idea of elite? Perl will get you maybe the cure for hiccups. You want us to let you in on the cancer vaccine, you better start studying Lisp.

Re:Not the interface

[TheNetAvenger]

Time Machine's ability to simply browse backwards through time in the folder, whilst still having the folder functionality usable is far beyond BackupPC.

I agree it is not just Backup software.

However, this is ALSO why 'previous versions' in Vista is more than a snapshot/backup interface as well. In Vista you can view folders as they exist at any previous time and even drag and drop the folder or files you want from the folder at a specific time in history.

Time Machine = 3D Interface of Files/Folders
Vista = Timeline List of Files/Folders

Time Machine = Uncompressed Backups to External Drive
Vista = Compressed Backups ANYWHERE + File Version Snapshots on main Hard Drive + Works on Servers and across networks (ie Can use Previous Versions on Folders/Files you have access to on Servers or other computers, and it displays that folder's snapshots and backups.)

Time Machine = Great Marketing
Vista = MS's Sucky Marketing

So Time Machine gets the cool buzz, when Vista is the cooler of the two technologies...

Re:Not the interface

[jav1231]

Hmmm I thought I saw him drag-n-drop it on the promo video. I don't currently have Time Machine turned on so I can't verify. OTOH, it's hardly like this guy's argument still holds much weight. Not a European Swallow, that's my point! :p

Re:Not the interface

[bodfa]

Actually with BackupPC you can look at a specific directory, and see all files that have existed in the dir. Even better you can see what all points in history the file has changed.

See if these screenshots help:
http://backuppc.sourceforge.net/info.html#screenShots

I also love BackupPC because I have several different OSes on my network and BackupPC can handle them all.

Just set it, and forget it... well... then check it later and see if it still works... Oh, and maybe setup some kind of off site system. That isn't quite the same slogan.

Re:Not the interface

[Anonymous Coward]

Hey, shouldn't you choose latin "per se" or Portuguese "por si"?

Re:Not the interface

[Anonymous Coward]

PER SE, not "per say"

Day-um

Re:Not the interface

[TheNetAvenger]

People are actually using vista/longhorn/2003?

Ya, in fact more people are using Vista than all the Macs users combined. And Vista didn't quite meet their sales expectations. Strange uh?

http://blogs.zdnet.com/microsoft/?p=598
BTW Old Article link, last numbers I saw Vista sales were close to 100 million.

People should note this when they think OS X is taking over the world, it is little more than a 'press/marketing' thorn to MS, and even if the Apple user base doubles 3 or 4 times, it would still be irrelevant to MS...

Re:Not the interface

[torstenvl]

1) All caps is shouting. I think you meant to use quotes.
2) It's spelled "goddammit." Goddamnit is a 1998 album by the Alkaline Trio.
3) We don't start sentences with "And"
4) Gerunds are nouns. Nouns require determiners. You mean "Your saying..."
5) All caps is shouting. I think you meant to use the emphasis tag. "...is somehow confined..."
6) Time Machine is non-intrinsically confined to the Finder, by virtue of non-power-users not knowing how else to use it. In addressing the question of whether it was, in fact, "not confined to the Finder per say [sic]," you failed to address the core issue of whether there actually might be some non-intrinsic confining of the use of Time Machine by design or circumstance. For this, you fail.
7) Denying the truth of X "per se" does NOT imply a non-intrinsic truth of X; it merely suggests it. You should learn the difference between "imply" and "suggest" ... unless you just like to use big-person words you don't understand.
8) Your phrase structure is awkward; it wouldn't "imply that either {X} or you just like to use...." It doesn't imply anything about his likes or dislikes at all. It may indicate them, but it does not suggest them, and it certainly, by no stretch of the imagination, implies them.

Please, think about what you're saying!

Re:Not the interface

[abhi_beckert]

The "interface" also includes things like automatically beginning a backup as soon as you plug in your external backup drive without any notification to the user (except for chasing arrows next to the icon), and pausing the backup if you un-plug it while time machine is running, again without notifying the user, to seamlessly resume where it left off next time you plug the drive in.

It's also using a new filesystem logging tool (called FSEvents) to fetch, instantly, a complete list of every single file you modified since you last plugged in the backup drive 6 days ago (which is quite an acheivement, considering the log doesn't take up several GB of space nor does it slow down the kernel).

Also, time machine is very intelligent about it's usage of space (within reason, it could be improved by slowing down the backup). Plus when your backup disk runs out of space, it gives you intelligent options for what parts of the backup to delete.

rsync is a wonderful tool, I use it several times a day and love it. But for backups it's nothing compared to time machine. rsync isn't going to run automatically if you have a drive that's not always plugged in (laptop? or maybe you keep your backup at an external location so it can't be stolen?), and it's going to harass you if you try to unplug the drive or turn the machine off in the middle of a backup, and it's going to take an hour to do a backup that might take 45 seconds in time machine (checking the modification date of 5 million files, then comparing each one with the modification date of another 5 million files takes a while), and you're going to run out of space in about 2 days if you decide to do *hourly* backups of your *entire* hard drive, which is what time machine does if you leave the drive plugged in permanently.

Also, whenever you install the operating system (either because your hardware failed, or you hacked the OS and broke something, or even if you just bought a new mac), you have the option of plugging in your time machine backup hard drive and it will skip the entire setup procedure, booting you strait into the same machine you left off with, all your software already installed and all your documents where you left them.

Re:Not the interface

[nschubach]

IMO, it is not the _interface_ that is cool about Time Machine, but the ease of use and the fact that it is fully automatic.

What's the difference? The interface is how you use software. If it's easy to use, it has a good interface.

If we are talking absolutes here, the most fully automatic application would have no interface. I'm not pointing this to you, but the combination of your post and the GP.

Re:Not the interface

[DaggertipX]

The backups are actually stored as plain files on your backup drive, so there really is nothing preventing you from dragging the file to your desktop or wherever, if you know what directory it's in. True, you wouldn't get to see the starfield flying behind the window while you did it, which is quite a tragedy...

Microsoft's implementation is useful, if you have the OS version that has it, and the savvy to use it. Apple's implementation is useful, if you have Mac OS.

Re:Not the interface

[Ant P.]

GUIs exist for GnuPG, but their installation is complex. $pkgmanager --install kgpg is complex? If that's too hard for you I heard Ubuntu has a GUI for the package manager too, preinstalled so you don't have to think at all!

Re:Not the interface

[Anonymous Coward]

Mac has about 6% of the market. If they doubled their share 3 or 4 times, they'd have 48% or 96% of the market.

Moderators: Please look at parent's posting history in this thread, and overall. He is obviously a troll.

Re:Not the interface

[COMON$]

What I dont get is how everyone seems amazed by this time machine idea. MS has been doing it for years with Shadow Copy, the exact same thing. The only difference is it wasnt integrated with the desktop until vista. But no one says a word if MS does it, where as if Apple does something it is touted a marvel.

It is just weird to me that on /. of all places no one has mentioned that MS has been doing this, must be for fear of retaliation from the linux geeks.

Re:Not the interface

[slashflood]

Time Machine's ability to simply browse backwards through time in the folder, whilst still having the folder functionality usable is far beyond BackupPC.

Not true. You can browse the directory tree and then select the date. The interface presents the state of the directory depending on the date. It doesn't show fancy icons, but the workflow is the same.

Re:Not the interface

[JanneM]

All in all, it is nifty, automatic, and relatively painless. Still waiting to discover the quirks It's right in the Ars review - it's hard link based, so it only works on the level of whole files. Any change anywhere in a file (like an email repo or database file) and the entire file needs to have a new copy. So for some users this will fill up the backup media with new copies of large files in a matter of days or weeks.

Re:Not the interface

[Emetophobe]

That makes it a cool-as-shit-goofy-3d-zooming-thingy that's also functional.

It's a UNIX system! I know this!

Re:Not the interface

[3choTh1s]

TM aggregates changes and you can roll back to any point in time.

No not really. TM does not aggregate changes in the way aggregate means. It determines if there is changes to a file and provides a hardlink if there is no changes. Otherwise it either deletes the file from the copy it produces for that time or it copies another complete file to the backup.

Think of it as CVS. It tages backup times but actually only copies new data as it's checked in.

CVS and SVN will only backup changes to what's stored reducing the the need for a whole bunch of disk space when things don't change that much. But in TM if you change a small thing in a big file, you're gonna need roughly twice the space of the big file to store it and it's backup, not to mention each time you make a small change you'll need even more space.
The rest of what you say is spot on. The interface is great and lots of applications(mostly those Apple-made) have direct integration so it's easy to find what you're looking for and get it back from the depths of file death.

Re:Not the interface

[pebs]

The interface is the goofy 3d zooming through space view. The ease of use and he's referring to is that the incremental backups that are stored all appear like full backups from a file system perspective.

Well, the typical rsync backup with hard links give you "incremental backups that appear like full backups from a file system perspective." I've been doing that for years on my Linux machines and all I have to do is browse a folder that has all my backups in seperate folders sorted by date. I can then find which date I want a file from and just access it directly, copy it over if I want or whatever. So does rsync w/ hard links have the same ease of use?

Personally, I think the cool and innovative thing about Time Machine is it has an API for integrating into applications. Other than that, its the same shit I've been doing in Linux for years. Only now, non-techies can use it easily.

Re:Not the interface

[Americano]

From your linked blog:

With Vista, "We've eclipsed the entire Apple installed in the first five months of sales," Kevin Turner, Microsoft's Chief Operating Officer, told attendees of FAM.

I find it amusing that sucn an insignificant thorn to MS rates a comparative mention... "We've totally outsold that dinky 5%-of-the-market guy." Why even bring them up or mention them if they're such an insignificant portion of the market?

even if the Apple user base doubles 3 or 4 times, it would still be irrelevant to MS...

Let's run some numbers. First, Apple currently has about 6% of the market, based on the most recent numbers I've seen. Second, I'll be generous and assume you meant "even if the Apple user base grows to 3 or 4 times its present size," since what you actually wrote indicated that Apple would grow to 48 or 96% of the market.

Now, do you *really* think that Microsoft wouldn't take notice if 18% - 24% of the desktop market was owned by Apple? That they wouldn't notice that 1 in 4 or 5 desktops are Macs? I'd humbly submit that if they disregarded that, it would constitute criminal disregard for their fiduciary obligations as a publicly held company!

No, Macs are not about to take over the world in the next 6 months. But they (or someone else) will manage to do so at some point in the next few years if Microsoft ignores those "insignificant" upstarts that only hold a few percent of the market. Happened to IBM, didn't it?

Re:Not the interface

[jacksonj04]

Shadow Copy is no use at all if your entire disk drive comes down in balls of flame, or if you don't have Premium or Enterprise, or if you simply don't know it's there. The first one defeats the purpose of a backup, and the second two are true for the majority of Vista users, never mind Windows users as a whole.

I've used both, and Time Machine, whilst functionally almost identical to VSC, is a much better product. It took me attaching a disk and clicking one button to enable it, and the interface is frankly amazing. Interface being something both Shadow Copy and this Linux clone thing lack - and configuration/interface is where Time Machine wins out. I can actually see it seriously increasing the number of Mac users who make backups simply because they don't need to engage their brains, or hunt down a backup application, or remember to burn a CD every week. It is literally plug in a disk, and click "OK".

On the subject of mentioning who does what though... where are the people screaming at the developers to have some originality? If this concept had been released first as an OSS product and someone at Apple went "Nice, we should have one" then people would be up in arms!

Re:Not the interface

[ssstraub]

VSS on servers has pretty much nothing to do with Vista as a desktop OS. It's using the same technology underneath, sure, but you can open a "Previous Versions" network share folder just as easily from XP as you can from Vista.

Re:Not the interface

[brassman]

Also, whenever you install the operating system (either because your hardware failed, or you hacked the OS and broke something, or even if you just bought a new mac), you have the option of plugging in your time machine backup hard drive and it will skip the entire setup procedure, booting you strait into the same machine you left off with, all your software already installed and all your documents where you left them.

Just as a data point -- I noticed the "restore from Time Machine" choice on the Leopard install, and as my MacBook was getting a little crowded at 50GB, I bumped it up to 120 and used Time Machine to do the restore.

Hardest part in the entire process was finding the itsy Torx driver needed to swap the caddy onto the new drive.

Re:Not the interface

[ToasterMonkey]

The ease of use and he's referring to is that the incremental backups that are stored all appear like full backups from a file system perspective. Whoa there, that's something nearly every backup/restore client does. That's sort of the whole point of an incremental backup.
Basically, full(past) + incr(date) = full(date)
Therefore, most restore clients show you full(date) when your browsing your backups. That's nothing special.

The ease of use is clearly in Time Machine's setup, which has been referred to as "two step backup."

Re:Not the interface

[akita]

Hmm, no we don't.
It's either "per se" from latin or "por si" in portuguese.
I should know.

Re:Not the interface

[NMerriam]

What's the difference? The interface is how you use software. If it's easy to use, it has a good interface.

And there, in one comment, is summarized the entire misunderstanding in some parts of the technical community of why Apple is renowned for designing software that is easy to use. Copying aqua icons and making shiny reflective doodads copies "the interface", therefore our GenericSoftware is just as easy to use as a Mac now, right?

The user interface to Time Machine is, hopefully, something no user will ever have to see (though they'll probably want to).

The ease-of-use comes from the fact that the first time you plug in an external hard drive, the system asks "Do you want to use this disk for Time Machine backups?", and if you say yes, you never have to do anything else. You never have to launch Time Machine, you never have to open the configuration utility, you never even have to know it exists. It comes preconfigured with exclusions and inclusions of what 99% of users need backed up. That's ease of use, and it has nothing to do with the application interface, because the user never even sees the interface.

Now of course in ADDITION, the interface to TM is all gee-whiz and fairly easy to grok for even nontechnical users, so if they do need to restore something some day, it's clear what is being done and which particular bit of data is going where.

Being able to restore a deleted address from WITHIN the Address Book is ease of use, and that has nothing to do with interface, it has to do with imagining how a user will want to manipulate data and not forcing them to dig around on the hard drive to find the address book database file and replace it with the older copy (which of course will also destroy all the other changes that have been made to that file since it was last backed up, so now you'll have to manually reconcile the two versions of the address book).

Rsync with a starfield screen saver does not Time Machine make. Of course, Time Machine without deltas eats up disk space a lot faster than a good rdiff-backup or System Restore alternative. Every solution has its tradeoffs, unfortunately.

Re:Not the interface

[jimmyharris]

Database file, yes. Mail file, that depends.

Apple's Mail client uses a single file per message so there's no problem there. Microsoft's Entourage 2004 uses a single database for all messages but I believe 2008 is moving to a single message per file which will support Spotlight indexing as well.

Re:Not the interface

[NMerriam]

MS has been doing it for years with Shadow Copy, the exact same thing...It is just weird to me that on /. of all places no one has mentioned that MS has been doing this

No, Shadow Copy comes up every time Time Machine is mentioned. The difference being, of course, that Time Machine is useful, fast and well-designed, while Shadow Copy is a strange collection of abilities that are bolted on in random locations in ways that they never quite accomplish anything accessible to the non-sysadmin (which makes sense, as the whole technology was never intended for the desktop, it's all a huge kludge to workaround the even older ridiculous MS file systems that won't allow backups to be made of locked files).

Storing all those diffs in the System Restore area of the disk basically defeats the whole purpose of creating backups, and makes it so it's only useful for reverting files from minor user error, though there's no way to know if you'll be able to recover from any particular issue since the configuration of how and when it copies what and how it chooses to delete old versions is a mystery.

So, to deal with all THOSE limitations, instead of fixing them, they included a whole new system backup utility to do REAL backups to separate media, and then made THAT useless by making the configuration abilities a complete joke, so unless you want to create an image of your whole drive and waste tons of space on the backup destination, it's pretty much impossible to be sure your important data is backed up.

Then, to top it off, you get to use this easy-to-read matrix and flowchart combination (because users love poring over product matrices) just to figure out which of the dozen different backup capabilities are allowed to be used in a given version of Vista.

So no, it's not the exact same thing. Copying data and making backups are nothing new for ANY computer company. The only amazing thing happening here is that a company finally made it not suck. Backup hasn't been this easy since the days when you could turn on the tape recorder and have all 64kb of your data on an easy-to-store cassette.

Re:Not the interface

[lee1026]

Note that the FOSS community is made out of more then one person. Different people can view the world differently and still part of the same group.

Re:Not the interface

[Divebus]

Something is fishy about all those numbers of market share and units sold. Lets see;

I know more people with iPhones (4) than use Vista (3) and they only sold 1.2 million iPhones in the U.S so far. 100 million copies of Vista? Maybe, but Idunno. Can't see them anywhere. Of the three people I know with Vista, two hate it and are ready to buy Macs. I know many many more people who used to own PCs who now own Macs (about 30 over two years). Seeing as one in every six laptops sold in the U.S. is a Mac, that market share claim is further in doubt.

Very few corporations have switched to Vista. There weren't any lines at Best Buy to get boxed Vista when it was released. Who's buying it? How many Vista machines have been rolled back to XP? How's XP selling next to Vista? Apparently enough to force Microsoft to allow it to be installed on new computers.

So, in the time Vista has been around, Vista shipments slumped in Q3, Apple has jumped a few percent in market share, Dell is selling Linux computers fairly briskly and Ballmer has been VERY noisy about little Linux blip on the radar. Something doesn't add up here - like someone is lying about it all.

Microsoft better watch their backs - looks like the user satisfaction figures are becoming leading market share indicators.

Re:Not the interface

[Daengbo]

Hardest part in the entire process was finding the itsy Torx driver needed to swap the caddy onto the new drive.

I'm not familiar with the Torx brand. The "itsy" thing kind fo goes against the oversized head trend. What kind of distance do you get from it? Does your caddy recommend clubs or just carry the bag? ;)

Re:Not the interface

[BrokenHalo]

Hmmm. Guess I'll find out how good the interface to Time Machine is as soon as I get past Leopard's shiny new Blue Screen interface.

*ducks* ;-D

Re:Not the interface

[Daengbo]

So I write in an extra right-click option on externally mounted drives to set it as the "Designated Backup Drive." I put the USB ID into /etc/dbdd.conf as"

# The Designated Backup Drive for the DBD daemon
ID 0718:0246 Imation Corp.

Then I start a daemon which uses HAL and DBUS to monitor for that drive being inserted, starting the backup process every time. Heck, I might even be able to do it with only DBUS and no daemon.

This sounds so trivial even I might be able to do it, and I haven't programmed anything since 1989 or so. If this isn't standard on distros a year from now, I'll be really surprised.

Re:Not the interface

[gerardrj]

Except it's more than that....

Show me your rsync solution run every hour and automatically manage disk space by only keeping the previous 23 hourly backups, 30 days of daily backups and monthlies for as long as disk space holds out.

When you're done with that make rsync work on an intra-file level so that I can restore individual records within a database structure from within the application managing the records (using the same exact interface as the file restore functionality). Example, restore a single calender entry from within the calender application.

When that's all ironed out make it all work seamlessly regardless of how often the backup drive is connected and turned on.

Re:Not the interface

[gerardrj]

The new Disk Utility in Leopard allows one to perform live partition modification, even on the boot drive/volume. Launch DU, shrink your boot volume, add a new one and set the new partition for TM backups. Viola, backups with only one drive.

That setup does nothing to protect against drive failure (it even causes more wear-and-tear on the drive), but it will protect against data corruption and accidental deletion/overwrite.

TM is not making disk images, it's copying files that have been changed since the last backup.

TM also works at a finer level than just files, it can restore individual records within a file (a single iCal entry, email message or address book entry for examples) without affecting the remainder of that database.

For the home/office user the hourly backups will also alleviate the "accidental overwrite" scenario: backup fires at 1am, you create a file at 11am, then accidentally save over that file with new content at 1pm then change the file three more times that day but don't realize all that until the meeting at 4pm. DV is useless since no version of the file is backed up. With TM the initial version and at least one of the revisions are in the backups. Simply fly back and grab the version (or versions) you need. Complete time to restore your data is probably 2 minutes. TM can't protect against all issues but it is quite a large safety net.

Re:Not the interface

[gerardrj]

Time machine will back up to almost any mountable volume (not those on airport base stations). You can have many systems backing up to a single server

Re:Not the interface

[NMerriam]

The beauty of Slashdot's level of nerdosity is that I can't even tell if you're being sarcastic :)

Re:Not the interface

[SanityInAnarchy]

I find that they aren't even mutually exclusive if you're talking about the same person.

Many geeks working on FOSS are much more interested in how well something works than how easy it is to get started with. This is why, for quite a long time, Linux was hard to install, but ran well while it was up.

This goes for UIs also. As a user, I'm much more interested in how well something works once I know how to use it than how steep the learning curve is. To a point, of course -- I still haven't learned TeX.

So, as a sort of irrelevant example, while the Firefox and Konqueror logos may kind of look like the Explorer or Netscape logos -- all browser logos seem to have a globe in there somehow -- you still are going to be clicking on Firefox, or maybe "Firefox web browser" or "Web browser -- Firefox", and not "Internet Explorer".

Believe it or not, this actually a problem for some people. But there are all kinds of similar problems -- for instance, The Gimp, for awhile, was about five windows. This works fine if you've got a decent window manager and/or a second monitor / virtual desktop system, but for many people coming from Photoshop, it didn't work at all. (There are other things Photoshop has that Gimp lacks, but I think Gimp was actually more usable on multiple monitors until Photoshop gained the ability to pop out windows. But my memory is fuzzy on that.)

Apple seems to do a good job of making it visually discoverable, though -- something only rarely seen in FOSS. For instance, I was using WindowMaker for years before I tried OS X, and the window manager certainly existed before I touched it. So when I saw OS X's Dock, I immediately knew what it was. But here's the thing -- so did everyone else, or at least, enough to use it (sort of) without panicing.

Maybe not as well as they should, of course -- I know far too many Mac users who keep some 20-40 things in their dock, most of them running -- I'm not exaggerating! But OS X knows how to swap, so while it is kind of stupid, it's not critically stupid, like not knowing how to run an app in the first place (could be problematic in WindowMaker).

Re:Not the interface

[TheNetAvenger]

VSS on servers has pretty much nothing to do with Vista as a desktop OS. It's using the same technology underneath, sure, but you can open a "Previous Versions" network share folder just as easily from XP as you can from Vista.


True Windows 2003 server supports this, however not the backup tracking.

Also this is relevant to Vista if it is on your network (like at home), Vista machines serving files provide this feature to other Vista clients.

Re:Not the interface

[Anonymous Coward]

Don't blame me, I voted for Kodos.
I'm offended by censorship.
Patriotism is akin to racism.

Since having an opinion also is akin to racism you should be more specific and less stupid.

Re:Not the interface

[papercrane]

You get the gold star today. That was just too perfect.

Re:Not the interface

[BobPaul]

Except this is done from a file system perspective and not in some proprietary format readable only by some slow and painful restore client (I'm looking at you, WinXP Pro/Server 2003, as well as any consumer backup solution I've ever tried.) Being able to simply plug the backup drive into another machine and browse for the file I want through the finder without worrying about a backup client sounds pretty enticing.

Re:Not the interface

[COMON$]

"from any particular issue since the configuration of how and when it copies what and how it chooses to delete old versions is a mystery."

This is completely customizable actually, I have all my shadow copies set up on my network to work on defined intervals by me.

Time machine, while being interesting and a definate bonus for OSX is only available with a external drive that I am aware of, if one were to take that path, why not just buy Retrospect? maybe I will feel different when I actually play with Time Machine. I am assuming that with time machine that you still need to have an OS loaded before it will work.

Re:Not the interface

[NMerriam]

This is completely customizable actually, I have all my shadow copies set up on my network to work on defined intervals by me.

Shadow Copy is somewhat configurable by the UI on Server 2003, but on all versions of Vista it's a black box. There's no configuration interface other than turning it on and off on the "system protection" tab. You can only affect the schedule by running scripts with Windows Scheduler, you can only tell it to exclude certain files via registry editing. And as far as I can tell, there's no way to configure which files are removed first on 2003 or Vista, so you could well wind up in a situation where you have 12 copies of your IE browser cache and no copies of the excel spreadsheet you hadn't edited in a few weeks.

Time machine, while being interesting and a definate bonus for OSX is only available with a external drive that I am aware of, if one were to take that path, why not just buy Retrospect?

Time Machine requires that the data be written to any drive other than the one containing the data it's backing up. This is because Time Machine is an actual backup system, not *just* a versioning system. It is Shadow Copy, Windows Backup, Image Backup and System Restore all in one API and interface rather than spread across a dozen different feature sets. If you really wanted to back up to the same physical drive, you could create a partition and TM would use that, but it's a bad idea so Apple doesn't encourage it. Restrospect is just a backup program with a scheduler, the same as 99% of backup apps out there that nobody uses because they're too inconvenient and complicated. That's not at all what TM (or for that matter, Shadow Copy) is or is trying to address, though TM does make backup programs with schedulers pretty much obsolete for the typical desktop user, while you'll still need to use a traditional backup program on Windows.

I am assuming that with time machine that you still need to have an OS loaded before it will work.

The Install DVD for Leopard is an OS as far as the system is concerned. Once you've booted from your the DVD, you can restore your backup, no intermediate installation necessary. You can do a bare metal install and restore your OS (for disaster recovery), or just have it restore the user accounts and data (for a new OS install, system migration, etc).

Re:Not the interface

[COMON$]

The Install DVD for Leopard is an OS as far as the system is concerned. Once you've booted from your the DVD, you can restore your backup, no intermediate installation necessary. You can do a bare metal install and restore your OS (for disaster recovery), or just have it restore the user accounts and data (for a new OS install, system migration, etc).

That is pretty cool.

Dont get me wrong, I am not knocking on TM or anything, I dont have the experience with it to really give it a good review, just an overview of the product. I am excited about upgrading my MAC users when the opportunity presents itself. All I am saying is that this really isn't new tech, Retrospect does a good job of this, bare metal backups, client side configurable console, etc etc. I think what it is going to come down to is that TM is available to non business PCs. With shadow copy I am only interested in it from a network standpoint, haven't played with it on a stand alone vista box yet, all my boxes are connected to DCs.

Historically, Apple has put out far more polished products than MS, I think from what you have said, the only thing missing from time machine is the ability to create a historical data section on the hard drive, much like shadow copy does, for users who don't have backup drives. But then again maybe apple is just encouraging good habits, now if they could just encourage people to rotate backups off site :)

oblig

[SailorSpork]

...But will it run on Vista?

Why not a simple SCCS?

[dpbsmith]

What I'd like to see is a very simple source code control system, built on the same design. Perhaps one that would just serve the needs of a single programmer.

The essential thing is that it should look like a file system, with direct access to the project directories at any state in development... write access to the current version, read-access to previous versions... directly accessible to any piece of code via the normal file API.

There should be no need for copying files back and forth from a central repository to a working directory.

It should be equally friendly to text and binary files. It should not take much disk space to store versions of files that have not changed at all from one project version/label/whatever to the next. It is not necessary or desirable to store just the diffs between text files; in the year 2007 we really can afford the disk space to store an entire new source file even if only a few lines in it have changed.

It should not rely on some central database that can be a central point of failure if it gets corrupted.

It should reliably serve both the functions of version control and backup. Bells and whistles in version control are less important than backup. In particular, if it's on an external drive and the CPU fails, you should be able to plug that external drive into a new CPU and go on accessing it immediately.

To those who work on hundred-engineer projects that need full-bore version control and CASE tools and so forth, peace. I'm not talking about a one-size-fits-all solution. I'm talking about a lightweight, simple, minimalist tool that as far as I know doesn't really exist today.

Re:Why not a simple SCCS?

[Anonymous Coward]

sounds like Rational (now IBM) Clearcase :
http://www.ibm.com/software/awdtools/clearcase/

Re:Why not a simple SCCS?

[corychristison]

I cannot agree anymore. I've done a lot of searching around (although, maybe not as much as I could have) and I cannot seem to find anything of the sort. I was hoping there was something available in FUSE... but (last I checked) there was not. I am not much of a C/C++ person (HTML and tools pay my bills) although I do know a bit, I wouldn't have much of an idea where to start with something like this.

Please, if anybody knows of anything that works like this in Linux, please let me know!

Re:Why not a simple SCCS?

[ajayrockrock]

What I'd like to see is a very simple source code control system, built on the same design. Perhaps one that would just serve the needs of a single programmer.

How about mounting a webdav file system with a subversion backend that has autoversioning turned on? That way, every time you write to the filesystem, SVN will make a new version. I did this for an office file server and installed track to point to the same repository. So now people have a cheap web interface to view revisions of documents.

http://svnbook.red-bean.com/en/1.2/svn.webdav.autoversioning.html

All the Mac's in the office mount the webdav repo, my linux box mounts it via fuse, and even windows has "web folders". It was kind of a fun project that turned out to be pretty useful.

--Ajay

Re:Why not a simple SCCS?

[hey!]

Well, sure. But why make something look like a file system. Why not have the file system do this for you, if it is keeping a journal of changes anyway for recovery purposes? Or if you're only journaling metadata, why not use the 30-60% of free space to store old versions of blocks, throwing them out with some variation of LRU?

Such a system would be useful but it doesn't, in my opinion, take the place of either source code control or backups.

If you're hard disk fails, you are screwed, so it's not a backup.

If you change your copy of the source so that it breaks everybody else's code, nobody knows this, so its not a source code control system either. Nor does it help you set up version 3.2.5 on different machine than the one that hosted it. And if you add tagging, branching and patching functions to it, it's not so simple anymore.

Still, it's a good idea. I've said for a long time that people ought to use SCCS functions for any set of documents that is important. SCCS provides both document management and some level of backup (although offline backup is still needed for the repository). Programmers have the advantage of working in documented, non-proprietary formats, so that it is possible to diff things meaningfully, but even so SCCS would be useful for just about anybody, if they can be trained to commit with the right frequency. Taking that decision out of their hands is better than allowing them to make really bad decisions, but it's not ideal.

If you are a single programmer, there is no reason not to use an SCCS. You don't have any special requirements, other than overcoming the sloppiness that besets even the best programmer's solo work. So if you are a solo programmer, I'd urge you to install an SCCS, and not rely upon any built in OS versioning.

Re:Why not a simple SCCS?

[MyDixieWrecked]

Although what you describe sounds pretty nice and is great since it would have zero learning curve, the best part of your standard source versioning system (svn/cvs) is that you take notes when you commit your changes and your revision #s contain groups of changes.

Your system would make it difficult to figure out what changes were made at what points in time and what changes should be grouped together.

It would be nice to be able to roll back to any state of any file in conjunction with svn, though.

Re:Why not a simple SCCS?

[pa-ching]

Most of your requirements could easily be covered by git (or perhaps another DVCS). It takes very little disk space; doesn't store diffs; has great version control, file integrity checking, and backups; repos are self-contained and easily moved from computer to computer.

For coding use-cases or the one you mentioned--writable working tree and read-only history--it's perfect. However, there are design tradeoffs in git that may not make it *completely* suitable. Its handling of large binary files is probably a little slow. You could probably lower the zlib compression to get that working faster.

As for filesystem integration, Resier4's plugin system or something similar is suitable for the job. Unfortunately, reiser4's future is uncertain...

Re:Why not a simple SCCS?

[Anonymous Coward]

I thought the same and wondered why no-one seems to have already thought about it.

How about be a virtual file system driver that intercepts the actual write calls and stores only a diff and time stamp to the actual file. To actually read older versions and deleted files you'd need certain tools but that's what you have to expect I guess. It could sit on top of a normal filesystem and would be quite seemless in integration. It certainly needs a higher specification but why has no one ever done this before? Am I overseeing something?

Re:Why not a simple SCCS?

[NullProg]

Linux Journal had a article on doing this back in 2001. The benefits are replication, history, and backups.

http://www.linuxjournal.com/article/5976

Enjoy,

Re:Why not a simple SCCS?

[DragonWriter]

It should not rely on some central database that can be a central point of failure if it gets corrupted.

You can't avoid some kind of repository. Even if the repository is just a directory in a regular filesystem that stores, for each version, the whole directory tree and with copies of any files that have changed, and special files somewhere in the structure identifying the relationship between versions and listing files that have been deleted, that is still a central "database" that, if corrupted, is a source of failure.

(Incidentally, I think something like that is the obvious structure to use to implement what is practical out of what you are describing; it has all the features you want other than the absence of a central repository that can be damaged, and its repository is no more subject to damage than any other part of the local filesystem. And it seems to me -- off the top of my head -- that it should be pretty easy to implement mostly as a usermode filesystem with maybe a handful of management utilities.)

Re:Why not a simple SCCS?

[p0tat03]

That wouldn't work very well for code... One important thing in any working repo is that all commits don't break the build. This also means work-in-progress files need to stay very far away, and one of the major reasons to use Subversion in the first place (atomic commits). I save a lot while coding, a habit from my old days of hacking on old, crappy machines that would give out suddenly without warning (Win98 boxes, what can I say). I'd hate to have my code auto-commit every time I hit Ctrl+S.

Re:Why not a simple SCCS?

[linuxrocks123]

Very cool idea.

I think it might be possible to use FUSE to create something like that over a normal CVS / SVN/ GIT repo.

Just a few months ago, I created a gigantic SVN repo of my entire home directory. It's been working better than I thought it would, but being able to use it from the file system would be ideal.

Re:Why not a simple SCCS?

[bgspence]

I liked Patch/Merge. I used it in the late 60s / early 70s on Burroughs Large Systems, B5000s and B6000s.

  It was not a file backup managment system, but a change managment system. It applied punch card code line changes to source disk files.

Each 'patch' was separately identified and could span files. Patches were named and numbered to see in a source listing what the last patch modified the source in a source flie printout.

It's real power was backing out old changes. You could backout or modify a patch made 10 changes ago, leaving the last 9 updates intact. Each functional update to a system was independent of the others. You did need to deal manually with any flagged overlaps.

It provided a merged view allowing you to drill down in time against lines of code to see how things evolved. A bit like Time Machine in its granularity and time history zooming. But, with punch cards and printed merged listings.

Wombat

[Anonymous Coward]

I've used Wombat in the past. It's basically a perl wrapper for rsync with a scheduler built in for hardlinked snapshots. Each image is a "full backup".

Ubuntu has built in backup system

[Colin Smith]

"Simple Backup Suite". Not quite Time Machine, but very simple and effective.

apt-get install sbackup

Re:Ubuntu has built in backup system

[Anonymous Coward]

apt-get install sbackup Come on now, don't make Ubuntu less friendly than it is. Just to to Applications->Add/Remove and search for sbackup. Click and install.

Longer than a command-line but it's all GUI and friendly.

Re:Ubuntu has built in backup system

[Anonymous Coward]

Maybe he's using debian, you insensetive cold!

Re:Ubuntu has built in backup system

[matt_martin]

Argh, if sbackup would just clean up after itself, life would be much better...
(Still doesn't remove old backups correctly in "logorithmic purge" mode)

Re:Ubuntu has built in backup system

[J0nne]

Agreed. I wish it was installed by default, so it's more discoverable. They could also make it a bit easier to recover old files, because opening >16GB tar.gz files tends to take a while.

Time Vault?

[MMC Monster]

Any idea if these guys are working with the Time Vault ( https://wiki.ubuntu.com/TimeVault ) people?

Re:Innovation

[sammyF70]

sigh .. yep .. i wonder when linux will eventually catch up with Apple and use something like "spaces" .. i mean ... wow! Those apple guys really know their shit. Think about it! .. virtual desktops?? wowwww man!!! wooooowww!!! That's innovative!!!

We *REALLY* need that for Linux!

Yeah, makes me wonder ...

[dazedNconfuzed]

I just drag my important files onto an external drive.

The whole point is that you don't have to do that, it happens automatically.
AND it catches all the files that you didn't think were important, but are.
AND it lets you roll the system back to the state it was in at any given time in the past (hence "time machine").
AND it takes care of any problems that can happen during backup (like "disk full", "power failure", etc.).

Re:Yeah, makes me wonder ...

[moosesocks]

I'm also going to go out on a limb and say that most users don't know where their .MBOX files are, or even what they are. But they'll definitely be missed in the event of a crash.

Or another scenario that's a bit more likely (especially with email inboxes it seems), the mail database gets corrupted, and before you realize it, the automatic backup overwrites the good copy on your backup disk with the corrupt one. I know of a few people this has happened to.

Time Machine is a very good thing, and I commend Apple for it, especially since their old backup app sucked, and wasn't even included in the OS.

Now, how about getting network backups to work properly, and patching Time Machine to gracefully deal with large files?

Preposition Nazi

[Anonymous Coward]

I think you mean you built it *on* rsync.

"something like"=/=real thing. technology missing

[wouterteepe]

Claiming to have created a backuptool "inspired on" time machine obliges one to give some more documentation than... ehm... none. A few things I know about time machine, which are not trivial: Every backup in TM is a fully consistend directory tree for which no special software is needed to consult it. Disk space is saved by using hardlinks on the filesystem in a very delicate way, including hardlinks to directories (!!!). As a result, one can very selectively delete backups without corrupting anything. (e.g., you don't want to know the state 11 am, but do want to know the states at 10 pm and 12 pm? easy facilitated without any special software). TM uses a special feature in Leopard to keep track of modified files and directories, in such a way that TM itself does not have to scan for modified files, but is informed by the OS of modified files. This notification does not even require a deamon process. Now I do believe one can wrap together something which does backups. But standard unix/linux tools don't offer the above facilities - AFAIK. And rsync certainly does not facilitate multiple hardlinks to a directory to be made. Therefore, this shameless plug probably does not offer something similar to time machine. Unless the author also claims that a Trabant is something like a Ferrari.

rdiff

[Anonymous Coward]

isn't rdiff more or less the same thing

this one has many advantages over timemachine

[Anonymous Coward]

for one, you don't have to have OSX to run it, another advantage is that you don't have to have apple hardware to run it. Two winning attributes if there ever was one.

Rsync backup is not Time Machine

[guruevi]

I have had this type of rsync backups for years now on personal computers and servers. There are several scripts floating around the internet that do exactly what Time Machine does. The problem is 1) usability and 2) interface.

No end-user is going to put an rsync script in their cron jobs and specify in what mounted partition to store it and then later use rsync to restore the specified files. -- if an end user understands at all what I just said of course

Time Machine's interface is revolutionary. It gives you a way of looking back in time at your own computer and does it in a fancy way consistent with the interface. It does so for any Time Machine enabled application including Mail, Address Book, i*. If you have to restore a piece of mail from backup I doubt you'll know the name of the file it was stored in rsync or any other type of backup let alone knowing how to restore it without removing all the new messages.

Why did we always have to be bashing users for not creating their backups again? Because it was too difficult and too time consuming to make them. Time Machine takes literally 30 seconds to set up and the rest is automated. That's why people will start making backups. It's not difficult anymore and it's going to save me a lot of headaches.

Just my 2c.

Re:Rsync backup is not Time Machine

[nagora]

No end-user is going to put an rsync script in their cron jobs and specify in what mounted partition to store it and then later use rsync to restore the specified files.

I do, and I do it on the Mac. I use linking to keep a month of backups stacked so that only the changes take up space but any particluar day can be restored in its entirity if needed. I got the instructions from ORA's Linux Server Hacks vol. 1. I also use a combination of Find and cp to keep 5 minute snapshots of all changes to important directories. I do that on Linux and OS X, and I've been setting that up for friends with Macs for years. It's no big deal.

I do agree, however, that if you use OS X you need some sort of good backup system because the hardware on Macs is appallingly poor quality.

TWW

Re:Rsync backup is not Time Machine

[deander2]

I have had this type of rsync backups for years now on personal computers and servers. There are several scripts floating around the internet that do exactly what Time Machine does. The problem is 1) usability and 2) interface.

i have too....i just built a GUI around it, similar to time machines. (w/o the 3d effects, but they're planned)

and i agree with you. that's why i built this.

Re:Rsync backup is not Time Machine

[Anonymous Coward]

I do, and I do it on the Mac. I use linking to keep a month of backups stacked so that only the changes take up space but any particluar day can be restored in its entirity if needed. I got the instructions from ORA's Linux Server Hacks vol. 1. I also use a combination of Find and cp to keep 5 minute snapshots of all changes to important directories. I do that on Linux and OS X, and I've been setting that up for friends with Macs for years. It's no big deal.

Oooh, you're so macho. If only everybody in the universe was as cool and 1337 as you, nobody would need Time Machine.

Back in the real world... I have the technical expertise to set up a bunch of hacky scripts like that. I don't want to. Nor do I think they would be an adequate replacement for Time Machine, which is likely to achieve greater coverage (TM can restore whole HDs) with less performance impact (TM uses kernel FS event notification so it doesn't have to do the moral equivalent of 'find' to figure out what has changed since the last backup) and less probability of bugs which will lose data sooner or later.

I do agree, however, that if you use OS X you need some sort of good backup system because the hardware on Macs is appallingly poor quality.

What an appallingly poor quality troll. The disks Apple uses are the same as anybody else's, OEMed from Seagate, WD, Hitachi, etc.

Re:Rsync backup is not Time Machine

[nagora]

Oooh, you're so macho. If only everybody in the universe was as cool and 1337 as you, nobody would need Time Machine.

This is true. Just going to the backup folder and finding the backup that you need and draging it to where you want it to be is a difficult chore for the point and grunt generation.

I have the technical expertise to set up a bunch of hacky scripts like that. I don't want to. Nor do I think they would be an adequate replacement for Time Machine, which is likely to achieve greater coverage (TM can restore whole HDs)

It could be done but no home user needs that or would want to spend the amount of storage it would require on it. Business users would certainly find it useful.

(TM uses kernel FS event notification so it doesn't have to do the moral equivalent of 'find' to figure out what has changed since the last backup)

I don't think notify is a winner UNLESS you are using the system on very large numbers of files. Since file/directory changes are cached, the find method is "fast enough" for any home or small business. In fact, it's very fast. But on a large file server or something where lots and lots of files change often it would fail eventually, I suppose, but then so might the notify method.

and less probability of bugs which will lose data sooner or later.

Well, after using the system for five years, and what with it consisting of about 10 lines of code, I'm pretty sure the bugs are all gone.

What an appallingly poor quality troll. The disks Apple uses are the same as anybody else's, OEMed from Seagate, WD, Hitachi, etc

Case design, on the other hand is terrible and heat issues abound causing all components, not just hard drives, to fail with disturbing frequency. On top of that, it's basically impossible to get a Mac fixed in less than three weeks unless you live in one of the two UK/Ireland cities that have Apple repair facilities. And even then you'll probably have to wait anyway while they source the part from the US.

TWW

Re:Rsync backup is not Time Machine

[HuguesT]

The quality of individual hardware components in macs varies a great deal, in fact. Sometimes it's top notch like the current Intel processor they use at the moment, but sometimes it falls behind, like right now the external 23" screen they sell for MacPros, which are both behind in terms of quality *and* much more expensive than their PC counterparts. The problem is that you can't pick and choose individual components if they are not to your liking.

Regularly Apple has a well-publicized problem with their hardware.

Apple is brilliant

[FranTaylor]

For stimulating the design of the new Linux backup system.

All linux users should tip their hat to Apple for renewing the interest in better backup solutions.

This is why free software rules.

And also why we need companies like Apple who raise the bar.

Re:Apple is brilliant

[stubear]

"For stimulating the design of the new Linux backup system."

"For giving Linux developers new ideas to ape because they were having to think for themselves and getting absolutely nowhere." There, fixed that for ya.

Re:Apple is brilliant

[bit01]

Grow a brain. Everything, including Time Machine, builds on the work of others. It's only "IP" fanatics who thinks that ideas come, de novo, from thin air.

---

Has your software been deliberately crippled?

Re:Innovation

[repetty]

> sigh .. yep .. i wonder when linux will eventually
> catch up with Apple and use something like "spaces"
> .. i mean ... wow! Those apple guys really know
> their shit. Think about it! .. virtual desktops??
> wowwww man!!! wooooowww!!! That's innovative!!!
> We *REALLY* need that for Linux!

You're not taking this well, are you?

erh?

[Anonymous Coward]

Sounds like a glorified trashcan.

Recovery tool is better than a backup tool

[emj]

It's more of a way to recover your backup tool. So you are right, Time Machine is nothing without the interface. It sucks not being able to recover data easily, and sadly most other tools seems to concentrate on snazy ways to backup, not how to recover.

Re:"something like"=/=real thing. technology missi

[Anonymous Coward]

Unless the author also claims that a Trabant is something like a Ferrari.

It is!

rdiff-backup

[JBv]

I use rdiff-backup with a cron job to do backups of my laptop to an external drive. It takes between 20-50 mins to do a daily backup of ~50Gb of data with about 200-700Mb of changed files. The only missing feature would be a "time machine" file browser in konqueror.

Re:rdiff-backup

[FranTaylor]

You need to reconsider the use of the word "only" in your description. Time Machine is waaay more than what you have there.

Re:rdiff-backup

[AusIV]

The only missing feature would be a "time machine" file browser in konqueror.

Meet Archfs. It's a FUSE file system that takes your rdiff-backup mirror and mounts it so that you can view a snapshot of exactly what your file system looked like after any snapshot was taken.

What rdiff-backup is really missing is an easy configuration utility. (And, if you're comparing it to Time Machine, some fancy effects while you search through old files).

For anyone who doesn't know, the first time rdiff-backup is run, it does a full backup of your file system. The next time, it uses rsync libraries to efficiently copy over any files that have changed. Then it diverges from standard backups - rather than keeping a complete copy of the old version of the file, it stores a (compressed) diff so that the old version of the file can be re-created from the current version. Consequently, you always have a snapshot of your file system in case a drive fails and you need to restore quickly, but minimal space is needed to keep older versions of files. Old versions of files can be recreated by applying one diff after another to reach the state of the file when it was first backed up. It creates a hybrid of a mirrored backup system and a differential backup system, while keeping disk usage low.

As I say, there's no easy configuration tool for it at this point, but I'd definitely recommend it to anyone who is comfortable with the command line and wants to be able to access any snapshot of their file system.

Re:rdiff-backup

[mjboyle]

FYI, I just saw Time machine do basically the same thing (~100 Gb of data with ~500 Mb changed) in under 2 minutes due to the FSEvents tracking that's been mentioned several times in this thread which allows it to avoid traversing the entire file system looking for changed files. This makes the hourly backup strategy practical. Even if you are on a laptop and have to connect up a physical external drive, keeping the backup time that short means that there are no excuses. I can easily tell myself that I can't afford to wait 20 minutes for a backup to complete, but I can almost always spend 2.

Also I think everyone should be congratulating Apple for making the FSEvents magic open to everyone, so if someone else wants to go and make a better mac backup program, they can get the speed benefit, too. I'm hoping that the updated version of SuperDuper will do this, leaving me with no excuse for not having both a versioned backup and a bootable backup updated at all times.

Re:Innovation

[Computershack]

sigh .. yep .. i wonder when linux will eventually catch up with Apple and use something like "spaces" .. i mean ... wow! Those apple guys really know their shit. Think about it! .. virtual desktops?? wowwww man!!! wooooowww!!! That's innovative!!!

We *REALLY* need that for Linux! Indeed. Especially seeing as Virtual Desktops have been available for Windows XP for over 6 years and they work on Win98 and possibly 95...

Meh.

[Enahs]

Give me the following:

1.) A backup system compatible with rdiff-backup, or at least as simple as rdiff-backup (i.e. the incremental copy, minus one folder, can be the latest snapshot)
2.) Make it work with gamin, so that backups are automatic
3.) Make 'thumbs' for files it understands, and store those as well

END RESULT: Who needs Leopard?

Re:Meh.

[FranTaylor]

Okay, one feature down, only 9999 more to go. I'm sure your Leopard-killer will be ready before the end of the universe.

Re:Meh.

[tomstdenis]

yeah because no other distro has [reading apple website]... a terminal, text editor, web browser, C compiler, etc...

right ...

Re:Meh.

[FranTaylor]

Yes, my wife uses all of those things on her computer. Not! Where is Garage Band? Are you telling me that a non-geek can compose music on Linux? Try explaining to your wife how to format and partition a drive in Linux. On the Mac you don't even need to explain it. I have linux on my desktop, but my wife cannot work with it. She needs a Mac. She is a massage therapist and has no patience for the computer crap that you and I put up with all the time.

Re:Meh.

[tomstdenis]

Suck donkey wang ya poser. Ubuntu was hardly difficult to install. Plomp cd in, click install, hit next a bunch of times, make a user account (via gui), reboot, voila.

Just because you're turd munching lack of knowledge rhetoric breathing no-brain loser doesn't mean the rest of us can't master the nature of "point and click."

Tom

Re:Meh.

[FranTaylor]

It must be REALLY easy to install if you can do it.

Okay, monkey-boy, but what are you gonna do after the install is over? Some of us actually do things with our computers besides install software on them.

Go ahead and compose some music with your copy of Ubuntu and get back to me when you are done.

I love linux. I use it all the time. I've been using it constantly since kernel 0.99pl12. It's on my desktop and my laptop. I install linux on headless servers with the serial port, in text mode, using PXE. Don't even dare to tell me how to install linux.

Re:Meh.

[tomstdenis]

It's funny you say that, because I *DO* compose music with my Ubuntu powered laptop.

Lilypond FTW!

Tom

Re:Innovation

[Cecil]

Or, you know, maybe it was just Time Machine that is ripping off Dirvish, which I've been using to do backups on my fileserver for years.

Restore support in Linux installer

[wumpus188]

Under Leopard, you can wipe your disk clean, put in Leopard DVD and reboot... one of the first options would be to restore system from Time Machine backup. With this tool, what is the point of including /bin, /usr etc. in the backup if there is no system restore support in Ubuntu installer?

Re:Restore support in Linux installer

[J0nne]

If you used sbackup, you can install Ubuntu normally, install sbackup, click 'restore', and it'll install all the packages you had, plus it will restore everything you chose to back up (if that's your home directory, this means all your settings). A 'restore function' on the liveCD would make it easier, but I don't see why you'd need to back up anything except your home directory (except perhaps some config files like xorg.conf, /etc/fstab & smb.conf, and /var/www if you were running a website).

Re:Innovation

[Anonymous Coward]

Indeed. Especially seeing as Virtual Desktops have been available for Windows XP for over 6 years and they work on Win98 and possibly 95...

I think sort of work might be more accurate

Re:Innovation

[corychristison]

Especially seeing as Virtual Desktops have been available for Windows XP for over 6 years and they work on Win98 and possibly 95...

Ooo! Is that when the screen goes blue because your system has been infested with pirates and you have to turn your system off and through it out the window to save your data? I loved that feature.

... Yes it's a joke!

more oblig

[russlar]

...But will it run on Vista? you make the assumption that Vista runs.

Not an rsync expert by any means but....

[antifoidulus]

One of the things that actually make time machine work well is that OS X keeps a log of every file updated on the system, and when the time machine daemon runs it looks at that log and knows which files to back up(as well as what time to mark them with etc). Now, maybe I'm doing rsync the stupid way, but doesn't rsync have to rescan every file on the system to see if it has changed? If you are backing up large directories that could be a large performance hit....

Re:Not an rsync expert by any means but....

[myz24]

No, you're correct. This is how rsync works but people like to brag about how they do things and don't want to stop and think for a minute that, Time Machine is indeed similar to other existing technologies but maybe, just maybe, Time Machine is marginally better than what they're doing.

Re:Not an rsync expert by any means but....

[Titoxd]

One of the things that actually make time machine work well is that OS X keeps a log of every file updated on the system, and when the time machine daemon runs it looks at that log and knows which files to back up(as well as what time to mark them with etc). There's one correction to be made there. Keeping a log of each updated file would be overkill, at least from a disk space standpoint, so FSEvents reports changes at the directory level instead. The rest of your point stands, though.

~~~~

R1Soft!

[zeeklancer]

rsycn is a bit slow, and I find that it can't cope very well with a large amount of files. The only Linux solution I have found to be anywhere near the capability of TM is R1Softs CDP solution. It is too bad it does not run on Apple, but it does run on Linux and windows. R1Soft does have remote backups which in my opinion makes it a much better solution all together. In any case, I use TM for my Apple boxes and R1Soft for Linux / Windows, rsync is way to slow on any platform. http://r1soft.com/ -Zeek

Re:R1Soft!

[Just Some Guy]

rsycn is a bit slow, and I find that it can't cope very well with a large amount of files.

Define "large". Gentoo's portage system is a directory with a sizable number of files, all synced against their servers with rsync.

Re:R1Soft!

[zeeklancer]

Really, when you get above half a million files rsync starts to choke, and requires a considerable amount of memory, which causes my server to swap out, thus trashing the performance of my server and causing the backups to take much longer than they needed to. With my R1Soft solution I am able to take backups on my busy servers on the hour, with little overhead. On my less buys servers with less power users I am able to take backups every 15 minutes. I think one of the big differences is the deltas are tracked in real time were as rsync will need to re-index on every backup. -Zeek

Re:"something like"=/=real thing. technology missi

[Bert64]

I believe linux does have facilities in the kernel to detect if a file has been modified... But i imagine you'd need a daemon to monitor this and store the list of changed files.
I wonder how long before linux does have something similar to time machine tho?

Re:"something like"=/=real thing. technology missi

[wumpus188]

This notification does not even require a deamon process.

Actually, TM uses FSEvent framework and does require fseventsd daemon.

Re:Innovation

[sammyF70]

Sure am ;) I just find that kind of "In the beginning there was Apple .. then everybody started copying it" Appleboy talk fun.

Re:Innovation

[sammyF70]

hmm .. used them for the first time in X-Windows on HP Unix Machines at the University ... around 1993.

Rsnapshot

[Urban Garlic]

When I first heard about "time machine", my first thought was that consumer-grade commercial software had finally discovered rsnapshot. It's packaged for Debian, and available in "sarge" -- that makes it at least three years old.

Rsnapshot is an rsync-and-hard-links based scheme that also doesn't store duplicate data, and provides nice date-indexed browseable full file trees, much like the way both "time machine" and this flyback gizmo are described.

I haven't been this excited since AOL re-invented "ytalk"...

Re:Rsnapshot

[Anonymous Coward]

"I haven't been this excited since AOL re-invented "ytalk"..."

That's funny. "ytalk" is a retread of the original "talk" :-)

Re:Rsnapshot

[kwerle]

Yup. I've been using rsnapshot on OSX for a couple of years, now. So my response to TimeMachine was "Nice. rsnapshot with a fancy UI."

I have not fooled with TimeMachine much at all. Not sure if it will replace rsnapshot as my "back up to a remote system" solution.

Re:Rsnapshot

[phoebusQ]

Except rsnapshot uses rsync, which must rescan the filesystem. With Time Machine, the FSEvents daemon makes that unnecessary in the common case. Also, Time Machine uses hard links to directories.

Re:Rsnapshot

[cycoj]

And your point is?! Not using hardlinks to directories might waste a few kB, big deal! The daemon might be an advantage, but I might not want a daemon wasting resources all the time.

FS with snapshotting

[CarpetShark]

Hmm. My understanding/guess back when I first heard about it was that Time Machine was going to use the snapshotting feature of ZFS. Other Linux filesystems do have this feature. It's new and cool, but it's not ultra-new or ultra-high-tech. And yes, version control has been doing something similar for a long time.

Re:FS with snapshotting

[krog]

Sadly, ZFS did not quite make it into 10.5 (ZFS read support is there, read/write support is experimental and present only in developer versions).

Re:FS with snapshotting

[Anonymous Coward]

Your understanding was incorrect. Time Machine was designed to run on HFS from the start; the idea that it needed ZFS snapshotting was invented by the intarweb. Filesystem snapshotting would be a more logical method of implementation if Time Machine was designed to operate within a single disk, but it's a backup program. You need to copy to a second device in order to backup.

And of course the technology isn't ultra-new. Duh. Study the history of computing and you'll find that nearly everything we do today was invented in some form before 1970. What's interesting about Time Machine is that it brings very slick automatic backup with time-based snapshots to any user who has the money to spend on a second disk. Linux filesystems with snapshot capabilities could be used as the backend of something like Time Machine, but that potential doesn't do anything for your average user. The new and cool thing is the frontend which makes these capabilities trivial to use.

Re:FS with snapshotting

[ToasterMonkey]

And yes, version control has been doing something similar for a long time. Uhhh... what, backups? Congratulations... yes, backup software and versioning systems share at least one common goal, point in time recovery.

Filesystem features like links, snapshots, folders, and more complex things like databases have all been used to implement backup solutions. Using filesystem features alone to track versioning info is not new, and _very_ limiting. Snapshots can be beneficial to a backup solution for things like block level incrementals. But, can you migrate parts of a snapshot set to different volumes? Let's say, move the older full snapshot to tape, and keep the incremental snapshots on SAN disk? Can you recover individual files from a snapshot without the full being present, assuming the full _could_ have been moved/deleted? Versioning systems can happily use filesystem features because they don't need to use multiple volumes.

I do hope that filesystem features catch up to the needs of modern backup systems, but even with ZFS, this doesn't seem to be the case. Not entirely anyway. Microsoft's VSS isn't there either (correct me if I'm wrong, but there's no way to pull a single file out of an incremental VSS snapshot on a remote volume without importing the whole thing into the filesystem right?). Backup software alone can still do more.

For the hundredth time, Time Machine is automated, very easy to setup, and very easy to use. It isn't doing, or claiming to be doing anything new or special in terms of backup method. Apple still did a great job.

Re:FS with snapshotting

[CarpetShark]

For the hundredth time, Time Machine is automated, very easy to setup, and very easy to use. It isn't doing, or claiming to be doing anything new or special in terms of backup method. Apple still did a great job.

You seem angry, friend. Don't read complete coverage into my posts about one particular aspect; it'll give you a hernia or something ;)

nice...but

[Marin3]

a pity it uses gtk, if it used qt it would be cross plataform, none-the-less great work :)

Another simple alternative

[pistk]

there is also another simple alternative: www.arkall.com It is an alpha version service, which is basically an online backup solution. The linux client is open sourced, and it has a nice fuse interface for linux and mac. It is still in an alpha stage, but if anyone is interested, you can request an invitation here: http://www.arkall.com/index.php?page=join

Breaks for me

[Anonymous Coward]

Traceback (most recent call last):
File "./flyback.py", line 813, in <module>
main_gui()
File "./flyback.py", line 738, in __init__
self.show_prefs_dialog()
AttributeError: main_gui instance has no attribute 'show_prefs_dialog'

Re:Breaks for me

[gearloos]

hmmm.. A program that would claim to a Linux version of Time Machine had better be pretty dang easy to use! If a user can't even install it without extensive knowledge of the packages on their system then this whole disscusion is moot! |null

Re:"something like"=/=real thing. technology missi

[HoaryCripple]

TM uses a special feature in Leopard to keep track of modified files and directories, in such a way that TM itself does not have to scan for modified files, but is informed by the OS of modified files. This notification does not even require a deamon process. Now I do believe one can wrap together something which does backups. But standard unix/linux tools don't offer the above facilities - AFAIK

inotify and fam provide this in linux.

What's wrong with sbackup?

[coolhaus]

For months, I've been using sbackup from the Ubuntu repositories, and it works just fine. It's a solid product of the 2004 Google Summer of Code. Why would you want to use some imitation Apple software when there's a perfectly good solution practically built in?

faubackup/dirvish do this

[CarpetShark]

dirvish and (my favourite, and the one I've tested) faubackup do this. In the debian package, you just edit the config file for how many days, weeks, months, and/or years of snapshots to keep, pick a (possibly remote) directory to put the backups in, and then add lines to a cron job for each directory to backup. Leave it alone, and whenever something goes wrong, you just go back to the date-labelled dirs and copy the files back.

Re:faubackup/dirvish do this

[dlsmith]

What you're describing sounds like a definite improvement, but will either of these tools automatically prune the archives based on disk space? That's important to me for two reasons: first (and less important), I don't want to be bothered to devise a cleanup scheme that will make optimal use of my disk space; second, an ideal cleanup scheme would be dynamic: if I make minimal changes to the disk for a few months, I should be able to hold on to more data, and if I suddenly write gobs of data to disk, I'd like to automatically throw away more old backups than usual to make room for the new stuff.

Re:faubackup/dirvish do this

[CarpetShark]

Yes. When you tell faubackup that you want X days and Y weeks, it keeps X snapshots for each of the last X days, and Y snapshots for each of the last Y weeks; no more. I assume dirvish works the same way.

Re:faubackup/dirvish do this

[CarpetShark]

Oh, you said based on DISK SPACE. Sorry, missed that. No, they don't. I do I agree: it'd be nice to just allocate a partition, and say, "OK, use that to keep my work as safe as you can." I guess it wouldn't be TOO hard to write a script that checks how much space backups are taking with faubackup, and to automatically adjust faubackup's config to not go overboard. It's also entirely possible to just delete any old copies that you don't want to keep from faubackup's store, as it makes hardlinks that act just like regular files. Not perfect though, no.

Bacula or EMC Networker or Veritas netbackup or...

[thanasakis]

I cannot understand how this is different from all these solutions. Invariably there is some pseudo-shell where you can browse through time, cd where you want, mark and eventually restore files.

Can anyone elaborate what is so different about this (except the claimed time machine likeness)?

I assumed a kind of simplified tape backup system

[simong]

Time Machine looks like a simplified interface to something like Netbackup or (more likely) Miranda to me. It's a handy thing to have natively but I was less than happy about how it works out of the box: in short it claims to take weekly full backups plus incremental changes every ten minutes. The weekly full backups are taken until the disk is full (Apple's words). I can see that a lot of people are going to get caught out by this: on my Mac Mini the full backup came to 44Gb, which on an empty 300Gb external drive (which seems to be about the most common size sold in Maplin and PCWorld these days) will come to five or six weeks of full backups.

There's no indication of what it does then: does it overwrite the oldest backup? Are you prompted to delete the oldest backup? Does it suggest you go and buy another drive? In an enterprise environment decisions would have to be made about what was kept and for what duration: you wouldn't keep everything indefinitely.

I'm going to have to see if it's possible to write some usable rules this weekend: off the top of my head, keep music and video and photos and the like backed up regularly and maybe only back the OS and applications up periodically, in the case of the OS particularly before the 10.5.1 update as I have bad feelings about it. I think there might well be some wailing and screaming from people who have just switched it on in a few weeks time when their external drive is full...

Hear Hear!

[asphaltjesus]

The major problem I have with distro installation CD's is there's no method of doing a system restore. Procedurally, it seems to me the process should go something like:

1. Delete everything in /usr and /bin
2. Reinstall everything in /usr /bin and
3. return apt to it's installation default
4. Do a diff on /etc and copy over the changed conf files to conf.old and then overwrite everything in /etc. modprobe.d needs to be moved before the new one goes in.

This is definitely one of those situations where it looks easy. I have no idea how much room would be needed on an install disk to do something like this. Much less if there's enough RAM disk to do the job.

Feedback welcome!

Eh...

[Per Abrahamsen]

> Yeah. There's more to Time Machine than just a one-off backup of your data. TM
> aggregates changes and you can roll back to any point in time.

Doesn't all backup programs do this?

Re:Eh...

[MightyYar]

Doesn't all backup programs do this? If you take "roll back" literally and don't read the rest of his comment, sure. But it's clear from the context that he means temporarily roll back and snag just one little part of the backup. Not only that, but from within any spotlight-enabled application.

I've seen clumsy interfaces that let you grab a backed up file from within a backup before, but never one that lets you preview - let alone do all this from within other applications.

Re:Eh...

[2nd Post!]

No, not at all. We are also talking about restoring systems. Not all backup systems like you restore per file independent of snapshots. Some systems only allow you to restore an entire image from last week (which may be several gigabytes in size) when all you need is yesterday's phone numbers.

Re:Eh...

[Per Abrahamsen]

The ones I have dealt with have all allowed you to retrieve a single file from within the backup granularity. Not sure what you mean by preview. Sounds like an interface thing.

The interface has always been horrible though, which kind of reinforces the original posters claim that the new (or at least less common) stuff is in the interface.

Re:Eh...

[Ziwcam]

> Yeah. There's more to Time Machine than just a one-off backup of your data. TM > aggregates changes and you can roll back to any point in time. Doesn't all backup programs do this?

I would like to clarify this a bit. Say for instance you delete 50 contacts one day, and 25 the next. A week later, you realize you deleted a contact that you needed. Most backup programs would only allow you to rollback the entire address book, re-adding all those contacts (and deleting contacts that you've added since then). Time Machine allows you to LOOK at your address book as you view progressively older backups, then when you find the contact you want to restore, to restore ONLY that contact (not the entire address book).

Re:Eh...

[Score Whore]

Enterprise storage systems have done this for years. They just don't market it as a consumer product because at their level disk space isn't cheap. Or just go download Solaris 10 or OpenSolaris and setup a ZFS filesystem. ZFS makes it very easy to build something like this if you care to. You need about one line of bourne shell to take the snapshot and probably a few dozen to clean them up over time or when disk space gets low.

When you want to get the old version of the file you're looking for, just cd into the snapshots directory and there is your entire filesystem as it existed at that time.

Re:Eh...

[shmlco]

Usually you get to a point when doing traditional incremental backups where you simply do a full backup and start all over again.

Re:Eh...

[TheNetAvenger]

Yep that is a sweet feature, and works the same freaking way in Vista.

Why on earth did MS's marketing not run with this feature and rename it 'super duper time traveler'...

Re:Eh...

[DaggertipX]

Because as is the case with a lot of things about windows, they don't seem to realize when they have a good thing. They wrote all the good backend code for a great feature, halfassed the UI so that the general populace wouldn't get it or want it, and then hid the entire shebang in the most expensive version of their software. Then again, there I go, thinking that maybe desktop users are MS's customers, as opposed to say - dell and hp.

Re:Eh...

[Anonymous Coward]

Yes. Apple's address book stores individual contacts in individual files.
(And Apple's Mail program stores each email in an individual file, etc. This way Time Machine doesn't have to backup a single 3GB mail file, only the new/changed emails, contacts, etc. It's a policy consistent throughout Apple's software.)

Re:Eh...

[Blakey Rat]

I always kind of wonder that, too. I used to think the reason is that Windows is sold mostly to corporations, and home users are kind of secondary to them (even though I believe their market is about 50/50.) But now that they've segregated the Home users from the Business users, they really could do some real selling of Windows features. Of course, with Shadow Copy there are two problems:

1) Shadow Copy and the accompanying Backup utility program only ship on Vista Business and Vista Ultimate. Why not for Home users? Or at least Home Premium users? No f-ing clue.

2) The Shadow Copy UI is pretty sparse. It's hidden in the Property pane of the file/folder you're interested in, which I'd guess the vast majority of users never see.

What's interesting about Microsoft's business focus is that it leads to some pretty cool features for home users, even if Microsoft didn't intend it that way. For instance, including Remote Desktop for free in Windows 2000 Pro, then upping that by creating Remote Assistance in Windows XP. That's another example where Microsoft did it first, and Apple's just now catching up.

On a less-related tangent, what's really bothering me about Apple lately is their hostility towards Microsoft. Apple and Microsoft used to have mutual respect for each other; you'd never see a Microsoft ad telling people how crappy Macintosh is, and you wouldn't see Apple advertising how crummy Windows is. Microsoft still follows the same policy, but Apple's entire marketing strategy now seems to be entirely based on telling people that Windows sucks. I mean, those Mac vs. PC ads, the inclusion of the BSOD icon for Windows servers, etc. Come on, Apple: Microsoft's doing good work, stop being such asses.

Re:Eh...

[pebs]

Usually you get to a point when doing traditional incremental backups where you simply do a full backup and start all over again.

Not the case when you use rsync + copy with hard links. Don't know if that's really "traditional incremental backup" but it might as well be considering how often its used for that purpose.

Re:Eh...

[MightyYar]

By preview I mean to see the contents of the file without actually "restoring" it. So you can know if you have the right version or if you need to go back/forward further.

Re:Eh...

[MightyYar]

I don't think that anyone's claiming that Apple is using some new voodoo or advanced technique to make Time Machine work. It's just that they are the first to make everything work together in such a useful way for the average computer user.

In fact, it seems like all that they are doing is making a bunch of hard links on an external drive - hardly rocket science. And yet, even though big external hard drives have been under $200 for years now, Apple is the only one to come up with a system this nice.

Of course, the only others even in the home game are Microsoft and Linux. Microsoft could barely get Vista out the door without removing most of the compelling features, and desktop Linux still seems hell-bent on copying MS - so they will almost always be a bit behind.

What's that, is my bias showing? :)

Re:Eh...

[DrgnDancer]

Err... The single most famous Apple ad ever was the "1984" ad which wasn't exactly polite to the competition.

Re:Eh...

[mdwh2]

I don't think that anyone's claiming that Apple is using some new voodoo or advanced technique to make Time Machine work. It's just that they are the first to make everything work together in such a useful way for the average computer user.

Vista has just the same ( http://www.engadget.com/2007/10/27/leopard-vs-vista-feature-chart-showdown/ ). Even ancient Windows 2000 has a backup that allows you to restore an individual file.

Also, "Time Machine" has limitations, in that it requires a second hard disk.

Just because you prefer Time Machine doesn't mean that it is better for everyone - and in certainly doesn't mean it was first! (This seems to be a common trend with the Mac - users think "It was the first for me" and therefore conclude it was the first for everyone.)

Re:Eh...

[jthill]

If Vista's actually does work that well, then you've asked a very interesting question.

Somehow, though, I don't think Microsoft suddenly woke up and figured out how to actually be fanatic about making minor hassles vanish. I'm betting Vista's feature works that well *after* you mutter the right keyboard and mouse incantation, with a lot of "well, of course"ing from people who just overwhelm them (the hassles) with competence, from the user side of the equation, and simply cannot comprehend the notion of doing it from the software end.

Re:Eh...

[NMerriam]

When you want to get the old version of the file you're looking for, just cd into the snapshots directory and there is your entire filesystem as it existed at that time.

Great, so how do you pull a single database record out of that file and merge it into the currently running database with one click?

Time Machine isn't the same as what "enterprise storage systems" have been doing for years. We're talking about a backup system that can do everything from bare-metal system restoration to bringing back a single deleted digit in a phone number, all with one interface so simple even your grandmother could use it.

Re:Eh...

[TheNetAvenger]

Shadow Copy and the accompanying Backup utility program only ship on Vista Business and Vista Ultimate. Why not for Home users? Or at least Home Premium users? No f-ing clue.


Vista Basic is crippled, but I don't think this is missing from Vista Home Premium (the standard version most OEMs use).

Ultimate and Business add an additional backup system for complete PC backup and restoring from DVDs, Nework Shares, etc, Vista Home Premium has the regular back up system like XP where you have to reinstall and restore.

I completely agree that MS feked up bad with XP and Vista. There never should have been a home version of XP, nor Home versions of Vista, Ultimate and Business editions only, with the only difference is what is installed on the default installation.

Re:Eh...

[MightyYar]

I think you misunderstand my argument. I'm not saying that Apple is first, second, etc. I'm saying that they are the first to implement a backup scheme that the average user will find useful and do so without creating any extra work. Extra work is what dooms backups.

Yes, you can turn on shadow copy. Yes, you can right click on an existing file and restore versions of that file. Yes, you can get most of this "for free" with ZFS.

But does the system automatically set this all up for you after plugging in an empty drive? Is the interface simple and intuitive, and is it available from inside of applications?

Time machine certainly isn't better for everyone. The geek who wants to have fine-grained control and know exactly what his system is probably not going to like it. It's not enterprise friendly since shares don't support it like Windows 2003 does. The implementation (being both backup and versioning at the same time) requires a lot of disk space. But for the typical home user? This is exactly what the doctor ordered.

By the way, you can make it run on a partition of your disk... it does not require an external drive. Of course, you lose the backup benefit, but it would work as a versioning system that way - albeit a wasteful one.

Re:Eh...

[Score Whore]

Great, so how do you pull a single database record out of that file and merge it into the currently running database with one click?

You can do this with one click for any database that runs on MacOS X, I'll give you $5,000.

All Apple has done here is added an API for the recovery side. When the app vendors implement it it'll be more impressive. But even that's not new. VMS (along with others) had file versioning built into the OS years ago. Personally, I'm more interested in a robust system than an API that is supported by a handful of apps.

Re:Eh...

[Blakey Rat]

I disagree. The 1984 ad didn't say "PCs suck, Amigas suck, OS/2 sucks" it said, "hey, Apple has something new and different." Apple might have been thinking "PCs suck, Amigas suck, OS/2 sucks," but the commercial doesn't *say* it, not like the current Mac vs. PC commercials which come out and say "hey Windows sucks."

Re:Eh...

[Blakey Rat]

Vista Basic is crippled, but I don't think this is missing from Vista Home Premium (the standard version most OEMs use).

It is. I had Vista Home Premium, and no shadow copy and no backup utility. I upgraded to Ultimate specifically to get these features. I believe they're even listed on Microsoft's site, although Microsoft's site is pretty vague about which versions do what: http://www.microsoft.com/windows/products/windowsvista/editions/choose.mspx

They have a feature for "scheduled backups" in Vista Home Premium, but it's much wimpier and less powerful than the "Complete PC Backup" feature, which is only on Business and Ultimate. (The Complete PC Backup feature can back up to a bootable disk image, as I understand it. I use Mozy.com for backups, so I haven't played with that feature yet.) That chart doesn't list shadow copy, unfortunately, but I'm nearly 100% sure it's only on Business and Ultimate.

Re:Eh...

[NMerriam]

You can do this with one click for any database that runs on MacOS X, I'll give you $5,000.

OK, I live in Austin, when do you want to come down?

Gee, "all they did" was add an API that lets any application restore arbitrary data from any point in time, with no impact on performance and no user configuration necessary. Plus, of course, all the system-level stuff that allows that API to work, which is completely new. But you're right, that's useless, I'd much rather be using VMS on my home computer. VMS can, of course, search all those backups for a single database record and automatically show you the most recent backup that contains that record, with one click, right? You'd rather have a robust system with basic features than a robust system with even more useful features? Interesting choice, I think it's safe to say you're in the minority on that one.

Wow, versioning was not invented by Apple!? Thanks for the newsflash! Next thing you know you'll say that search over a network was not invented by Google, therefore clearly everything they do is just a ripoff of the old Gopher directories that people compiled manually.

File transfer wasn't invented by Bittorrent, therefore obviously it's just some boring API that is ripping off "cp".

Perpendicular magnetic storage? Bunch of bullshit. You were able to store data on punch cards years ago, therefore anything that stores data is just some shiny object for newbies to get excited over because they aren't as smart as you are. All it does is store "more data" in "less space"!? VMS was able to store data years ago.

Re:Eh...

[TheNetAvenger]

Yep, just had a chance to sit down to a Home Premium system, and Previous Versions is not available. I am having techs check for a tweak that turns it on, as this is another reason why I hate the various versions of Vista. MS feked up big time with the home versions.

However, Home Premium does have basic backup tools like XP did, just no complete restore function, you have to reinstall the OS and then restore from the backup instead of it being an automated recovery.

MS just needs to kill the Home versions, offer free upgrades to Ultimate, and then only offer Business and Ultimate (since businesses don't want games, media center, etc. installed)

Thanks for fact checking this...

Introducing ext3cow! Time-Machine for Linux

[SpzToid]

from the website:
http://www.ext3cow.com/

Ext3cow is an open-source, versioning file system based on ext3. It provides a time-shifting interface that allows a real-time and continuous view of the past. This allows users to access their file system as it appeared at any point in time.

Ext3cow was designed as a platform for regulatory compliance, and has been used to implement secure deletion, authenticated encryption, and incremental authentication. See the publications page for more details.

Some advantages of ext3cow:

      1.
              It does not pollute the name space with named versions
      2.
              It has low storage and performance overhead
      3.
              It is totally modular, requiring no changes to kernel or VFS interfaces

Re:Introducing ext3cow! Time-Machine for Linux

[gearloos]

one word. Moooooo

Re:Introducing ext3cow! Time-Machine for Linux

[Tyr_7BE]

That's exactly what I was thinking when I read this. More like 'introducing ext3cow for Apple in the form of Time Machine'. I don't know how stable ext3cow is, or how much of a hack or work of art it is, but I do know some people who have been happily time shifting around for a while now. And this is with an actual filesystem doing your copy on write for you, not some python/rsync hack.

Re:Introducing ext3cow! Time-Machine for Linux

[Bigjeff5]

This looks to be very very similar to what the new Leopard file system is doing, only leopard seems to keep the changes in a "log" in each folder, where ext3cow makes tagged filenames (if I understand it right, I could be way off on that point). This ext3cow looks like it basically points changed data to the old file with the old data, not overwriting anything, and hides older versions of the incrimental data.

This will mean that the drive will slowly grow, but the incrimental-down-to-the-filesystem setup means very little extra data use unless you are completely re-building files constantly with completely new data, and snapshotting multiple times a day, or whatever.

It would be really easy I would think to make a frontend app for ext3cow to do the backups and be just as functional and easy to use as Time Machine.

Cool beans man!

RetrospeKt

[dsvilko2]

You can also take a look at my quick-hacked frontend for rsnapshot (for KDE) - RetrospeKt: http://www.kde-apps.org/content/show.php/RetrospeKt?content=57952 I wrote it mostly for myself but it is really handy.

Re:RetrospeKt

[tzanger]

Nifty little utility, I've installed it. However you are trying to use the "kdesudo" command, when it seems to be "kdesu".

Re:RetrospeKt

[dsvilko2]

Hmm, I have a kdesudo on my kubuntu...

Re:Warning! Do Not Use!

[Anonymous Coward]

Let's see who can come up with the most interesting way to complete the following sentence:

"That's not a gloryhole, that's a..."

No Apple Bugs For Linux!

[Anonymous Coward]

Please! Time Machine doesn't even bloody work! Or anything else in Leopard. I just got my money back from the Apple shop for that pile of shit today. Thank goodness for UK merchantable quality laws.

Re:No Apple Bugs For Linux!

[nikostheater]

Well,i am sorry that you are not satisfied with it,but my experience is totally different than yours.. I find Leopard to be absolutely fantastic.And it works.

What's so new about Time machine?

[MrSteve007]

I have an honest question that no one seems to have answered. What is so exciting about 'time machine?' It is being extolled as new and groundbreaking, but I don't see much, if any of a difference with the 'previous versions' option within Vista, or 'shadow copy' which can be used in a network environment with XP and server 2003.

Both of those windows-based solutions, which have been out for quite some time, allow you to restore an individual file or folder from a wide range of dates. My setup backups files at midnight and 9am everyday, and I can any version of a file going back nearly 3 months. If I were to reduce the backups to once daily, 6 months of version changes on each file is plausible.

example: http://www.steveallwine.com/images/previousversions.jpg

The only arguement I can find about why Time machine is innovative is comparisons between it and system restore on the PC. Since these are two entirely different functions, I don't understand why its brought up.

Re:What's so new about Time machine?

[bestinshow]

Err, if the files aren't changing, why is there a difference in the amount of backups you can have if you backup daily or twice daily?

There's plenty of information available on Time Machine, some in this story even, so you can easily find out what Time Machine has and why it is different to the offerings in XP and Vista. I'd guess a major one would be ease of use, getting more than 4% of users actually backing up their files. Functionality is pointless without accessibility and usability.

Re:What's so new about Time machine?

[phoebusQ]

1) Time Machine uses the FSEvents daemon to keep overhead to an absolute minimum 2) Time Machine uses file and directory hardlinks to minimize space taken, but provide a complete, usable, identical directory structure to the original 3) How to you do previous versions on a deleted file? 4) Time Machine provides a minimalist, automatic, and "easy-to-use" backup solution that people want to use. Making people want to use backup software is a coup in itself.

Re:What's so new about Time machine?

[bar-agent]

The biggest and most exciting thing to me about Time Machine is the plug-in system. Time Machine itself can just restore files. But the plug-in system looks like it allows the application to extract individual pieces of data from within backed-up files, and to treat a set of files as one unit for browsing purposes. Plus, there's the QuickView tie-in allowing apps to preview the contents of a backup.

Re:What's so new about Time machine?

[Spy Hunter]

Time machine backs up to a different disk, so it can save you in the event of a hard disk crash or other catastrophic event. Time Machine has a cool interface that is deeper than just a starfield wallpaper and some 3D effects. For example, you can take a spotlight search back in time to see what it would have found in the past.

In some ways Time Machine is not as cool as shadow copies, because it doesn't keep every version of the file with disk-sector-level copy-on-write; it does backups at specified times on its own schedule, and if a file is modified at all it backs up a whole new copy. If someone integrated shadow copies with the time machine interface, and also kept the features that back up to a different disk, then it would be the perfect all-singing, all-dancing backup tool.

Re:What's so new about Time machine?

[Anonymous Coward]

Ars Technica Review of 10.5: Time Machine

I never used Shadow copy, but my mother can and does use Time Machine.

Re:Why not a simple SCCS? (No use git)

[Anonymous Coward]

No, use git. see the Google Tech Talk: Linus Torvalds on git

ZFS automatic backups

[timf]

I've got something like this for Solaris/OpenSolaris users at:
http://blogs.sun.com/timf/entry/zfs_automatic_for_the_people

(pretty screenshots on the blog post for the previous version
http://blogs.sun.com/timf/entry/zfs_automatic_backup_0_1 )

  -- plug a USB disk into your ZFS-based OpenSolaris box, and any filesystems you marked as being interesting enough for backups will get sent to the USB disk. If previous snapshot streams are already on the USB disk, then we do an incremental backup. GUI notification via notify-send(1) Works for me!

This still work, specifically there's no restore GUI right now, and the configuration is a bit clunky, but still...

"per se" not "per say"

[Anonymous Coward]

"it's latin! Read a book!"

"sorry Andy"

Re:Innovation

[Hatta]

Virtual Desktops on windows don't really work. MSVDM crashes Vector NTI reliably. I have never seen nor heard of a linux app that crashes with virtual desktops, so I don't think it's the apps fault. MSVDM is buggy crap that's not even worth mentioning.

Re:"something like"=/=real thing. technology missi

[acvh]

Actually, Time Machine only knows about directories. If FSEvents tells it that something in a directory changed it then compares the live directory to its last backup and backs up the changes.

My biggest beef is how Apple defines "changes". TM will rebackup a file if the permissions change. RSync won't.

rsync vs. intermerzzo for making a "Time Machine"

[Anonymous Coward]

Anyone considering making a "Time Machine" for Linux should look into the work of the InterMezzo File System layer. It is not a classic file system due to the fact that it doesn't store anything itself. Instead it sits as a layer between the applications and an actual storage file system (such as ext3) to track all changes as they occur. While rsync must do a fairly heavy weight scan for new/changed files, InterMezzo is able to skip the scanning step due to having already tracked these files at the time they where written.

Re:Warning! Do Not Use!

[geminidomino]

Let's see who can come up with the most interesting way to complete the following sentence:

"That's not a gloryhole, that's a..." "...space station."

(Who didn't know that was coming?)

Your are right - and wrong Re:Not the interface

[Anonymous Coward]

Lack of easy to use, comprehensible interface does hinder all open source, not just Linux. I suspect however, that people do not spend huge amounts of time coding arcane items, reviewing the bugfix reports provided by their users and fixing problems that appear only on one user out of thousand's screwball setup, and then say, "To heck with making the GUI work." I suspect, rather, that GUI is equivalent to proofreading - anyone can tell when a document has been poorly proofread, but fewer good proofreaders will work for free than people will write for free on the Internet. Of course, good user interface starts with documentation, if the GUI designer can't tell what the program does, he can't make a GUI to tell the user.

Funny

[msimm]

I've been looking into this soft of thing on the related topic of replication which slowly found it's way onto my home system as a great backup solution (in that case PeerSYNC). When you think about it backup really *is* replication, and ideally real-time replication. Under Linux there are a couple of great places to start (rsync may be a good engine, but certainly not the start point unless you want a polling or on demand solution) namely inotify and on older systems dnotify. The idea being to mirror the file system (using whatever means suit) to [device] as immediately as it updates in order to avoid data loss.

The thing Apple seems to have accomplished is making smart backup more then just a smart backup engine including the client in the design, which really is pretty revolutionary.

Typo on the "About" Screen

[Anonymous Coward]

The word "loosley" should be spelled "loosely".

Not terribly important, but people will be more willing to believe that your code is fairly well written (and tested) if you've also taken the time to spellcheck.

Re:Typo on the "About" Screen

[Anonymous Coward]

I also found waldo! Made me shit bricks too!

Rsync Backup System

[spencerogden]

If you are interested in concept of using hardlinks to simplify your back experience, you can either set something up manually with rsync, or you can check out Dirvish: http://www.dirvish.org/ .

As other have pointed out, the pieces that are missing are:
1) No directory hardlinks, meaning you backups take up more space than you would like, since you have to create the whole directory structure, and then hardlink each file.
2) No FSEvents, so a full tree compare must be done by rsync each backup. So backups are not instantaneous, and are a bit intensive
3) No nifty interface

All that said, if you read the ArsTecnica article about how the Time Machine backups are stored, they are stored in a manner almost identical to Dirvish.

For me Dirvish makes it trivial to have a backup taken each night, with various expiration rules, and each backup being essentially an incremental backup.

Re:Warning! Do Not Use!

[QRDeNameland]

My first thought was...

"That's not a gloryhole, that's a Republican US Senator!!"

Then I thought...what's the difference?

OMG, too many tail lights...

[Anonymous Coward]

So let me get this straight:

Apple chases Microsoft's tail lights, by ripping off Windows XP's "system restore" feature and rebranding it "Time Machine".

So Teh Lunix, not wanting to be outdone (since chasing MS's tail lights is THEIR gig), Teh Lunix rips off Apple's rip off of XP. So does this mean that Apple's chasing of MS's tail lights is now obstructing Lunix's view of the Windows tail lights, or is it that Teh Lunix is now so far behind Windows 95 that they can no longer even see the tail lights, and have to chase Apple's tail lights instead?

This is just so damn confusing.

I suspect..

[msimm]

that the FLOSS community does (mostly) realize this. But I also suspect that good interface design is difficult. I used to feel more like you seem to, but after years of watching very smart ideas come together with very mixed results I'm beginning to think that there's a layer to sophisticated design that is very unique. Apple seems to consistently shine here and for whatever reason talented designers don't seem to be as keen on joining OSS efforts. Maybe programmers are more political then designers?

Re:Warning! Do Not Use!

[Anonymous Coward]

"That's not a gloryhole, that's mmmfrgl...gack..."

rdiff-backup

[introp]

I can't help but mention rdiff-backup, which stores deltas of files changes going backwards in time, allowing one to retrieve the file at time X with a minimum of storage waste. The interface is, of course, nowhere near as nice as Time Machine.

FSEvents shortcoming

[quacking duck]

There's a big downside to FSEvents' "folder changes only" model.

If you move a file or folder to another folder, or even just rename it, it will copy to backup rather than hardlinking to the previous location/name (I verified this using "ls -la" in Terminal and checking the hardlink counts). Best to avoid constantly renaming your video files ;-)

Now, I can see why this happens: although every file has a unique ID that doesn't change when you move it (this is why aliases will find their targets as long as it's on the same drive), FSEvents isn't logging these for TM to pick up. So, TM knows the folder contents changed somehow, does a folder diff, and copies to backup anything "new," and stops hard-linking any file that's been "removed" from that folder.

Hopefully FSEvents and TM will be updated somehow to handle moves and renames, but I suppose they've evaluated the performance tradeoffs already.

Time Machine has been around for a while...

[Pausanias]

At least since 2004. Just bear with me and take a look at the following HOWTO.

http://www.mikerubel.org/computers/rsync_snapshots/

I've up a system exactly like the one described above, and had it running, on Panther and now Tiger. The same setup can be used in Linux, or actually any posix file system that allows hard links.

I've been enjoying beautiful backups, with each subdirectory being a perfect image of my home directory at any given date, for about three years now. What does time machine do exactly that's different from this, other than fancy graphics?

Re:Time Machine has been around for a while...

[beh]

OK - there are a number of things -
if you use rsync for hourly snapshots to keep everything for the past 24 hours, you would probably do that by creating 24 separate target repositories (i.e. 24 full backups); OR, you allow rsync to create backups for your files - at most 24, but you purge any file older than a day (i.e. if there are only 2 versions for a day, the 3rd version older than 1 day gets deleted).

Now you want to extend this to keep ONE daily snapshot. Either again, create full repositories; OR you need to make your purging routines for numbered backups smarter in only purging all 'hourly' backups older than a day; but leave the last change for each day before in there as well.

If you have that, you have the FILESYSTEM side covered.

But, this does not allow you yet, to transparently go back in 'view only' to look at past events within your applications (this needs the application to support your backup tool, which is one of the reasons why a number of applications need to be changed for Leopard).
Take the earlier example of an accidentally deleted address book entry - which you can restore from within TimeMachine - simply start TM while your addressbook is the 'current application' (the one having input focus). In TM, you will see your open address book, going back in time it will show the state of the applications saved data 1, 2, 3 or more hours or days ago (depending on your selection). In iPhoto, it would show you the state of the Albums you had at the time; but neither of the two apps would allow you to SAVE back into the past. Once you find the record you like, you select it and click restore, and TM will go back to the desktop and the data you have selected goes back into your app (this can even be a PARTIAL restore of the data, say ONE of two albums you deleted).

In terms of Leopard compatibility and apps, most apps need updates, because they do not keep their application data in sync on the disk -- e.g. you could have a partially updated file at the time TM takes the backup, which might result in unusable data for that particular snapshot. I haven't looked on what exactly the app needs to do, but TM will interact with the applications directly to save the correct states; or to display the correct information if you go 'back in time'.

And this part definitely goes way beyond other backup solutions I have seen, because backups up to now only care about what the filesystem says; and do not care about integration with applications.

That said, what I do not like about TimeMachine is that it is restricted to one destination volume (my MacPro has about 1.7TB in disk space; there is no single disk that could take a full backup of everything. Luckily, I have a sizable amount of data that rarely changes - so I back that up manually and exclude this from TM. Otherwise I would basically need a separate RAID to put the backups on...

Re:Time Machine has been around for a while...

[Pausanias]

That's not true. Read the link I posted above. rsync can create hard links, through the magical --link-dest option. This means that you never need to do full backups, or worry about keeping track of several versions.

You can set up rsync to do hourly, daily, weekly, and monthly backups, all taking up minimal (and I mean minimal) disk space, using hard links.

All you have to do is rm -rf your oldest directory of each kind (hourly daily, weekly, monthly) every time you run.

Heck, I'll even give you the BASH script I use to do this. Call it via cron: e.g. snapshot daily

#!/bin/bash
# This script makes a snapshot of several directories using rsync.

# Version 0.5, Rob Bos, September 8, 2002
# Modified by Pausanias August 2004
# This script is released under the terms of the
# GNU General Public License, version 2.0.

# run "hourly" from a crontab; run daily, weekly, and monthly from /etc/cron.{daily,weekly,monthly}/.

[ -z "$1" ] && echo syntax: $0 snapshottype baselinesnapshottype && exit 1

type=$1 # hourly, daily, monthly snapshot?
logfile=/var/log/snapshot.log
backuproot=/Volumes/Sutton/ibackup
backup=`cat /etc/snapshot/include.txt`
excludefile=/etc/snapshot/exclude.txt

echo `date` >> $logfile

[ ! -e "$backuproot" ] && echo "$backuproot not mounted." >> $logfile && exit 1

# Assume that the baseline snapshot type is daily unless otherwise
# specified.
if [ "f$2" == "f" ]; then
basetype="daily"
else
basetype="$2"
fi

# if you run every two hours, make it 12, every hour, 24, etc. try to span the entire day.
# The below setting runs 8 times a day (every three hours), keeps 7 days, 4 weeks, and 12 months
# of backup.

hourlymax=8
dailymax=10
weeklymax=5
monthlymax=8

# This simplifies things down below so I can use $max to delete
# the overrotated snapshot, as well as $hourlymax/$dailymax to do
# rotation checks.
[ $type == "hourly" ] && max=hourlymax
[ $type == "daily" ] && max=dailymax
[ $type == "weekly" ] && max=weeklymax
[ $type == "monthly" ] && max=monthlymax

# Rotate the current list of backups, if we can.
if [ -d $backuproot/$type.1 ]; then
oldest=`ls -dt $backuproot/$type.* | tail -n 1 | sed 's/^.*\.//'`
echo Oldest is $type.$oldest >> $logfile 2>&1
for i in `/usr/local/bin/seq $oldest 1 -1`; do
mv $backuproot/$type.$i $backuproot/$type.$(( i + 1 ))
done
linkopt="--link-dest=$backuproot/$type.2"
else
linkopt=""
fi

if [ "$type" == "$basetype" ]; then
rsync -xva $linkopt --safe-links --delete --delete-excluded --exclude-from=$excludefile $backup $backuproot/$type.1/ >> $logfile 2>&1
elif [ "$type" == "daily" ]; then
[ -d $backuproot/hourly.$hourlymax ] && mv $backuproot/hourly.$hourlymax $backuproot/daily.1
elif [ "$type" == "weekly" ]; then
[ -d $backuproot/daily.$dailymax ] && mv $backuproot/daily.$dailymax $backuproot/weekly.1
elif [ "$type" == "monthly" ]; then
[ -d $backuproot/weekly.$weeklymax ] && mv $backuproot/weekly.$weeklymax $backuproot/monthly.1
fi /usr/bin/awk 'substr($0,length($0),

Re:Time Machine has been around for a while...

[beh]

Thanks for that, I really hadn't seen that option before... Nice!

(still - this only takes care of the file-system side of things; not the app-support around it; but, at least for my linux servers, I will have a look at that...)

Re:Time Machine has been around for a while...

[Obsidian Butterfly]

rsnapshot does this already.

Other Things Missing

[99BottlesOfBeerInMyF]

A number of people have pointed out some of the major deficiencies of this software in comparison to Time Machine. There are a couple of items, however, that no one seems to be mentioning and which I think will have some of the biggest, long term effects. First, Time Machine includes easy APIs so that other programs can access the stored data from within their application. Second, it is included in the standard install so developers can rely upon it being there.

Why does this matter? Think of all the applications in which versioning would be really nice, but it just isn't available. Your address book, for example can look up old contacts or numbers or addresses. Your development tools can automatically load an older, version of that code you're writing to recover that function you did not think was needed anymore, even if you did not write it to a versioning server. Your video games can take you back to older saved games or versions of characters before you sold that really cool item. Photoshop, Word, OpenOffice, etc. can use it to revert changes to a file all the way back to last month.

The difference is that while many users will never take the trouble to learn how to use a backup system and properly recover an old version of a file, they might trouble to plug in a Time Machine drive and then use the interface to backed up versions from with their applications. It seems strange that everyone is ignoring the cool new API for developers and concentrating on the integration in the finder, which will probably be the lesser used portion of Time Machine.

Windows, Linux

[Anonymous Coward]

"Me too!"
-_-

Re:"something like"=/=real thing. technology missi

[deander2]

Claiming to have created a backuptool "inspired on" time machine obliges one to give some more documentation than... ehm... none.

ok, fair enough. =P

to answer your questions, yes, i am creating a fully consistend directory tree for which no special software is needed to consult it. hardlinks are used, but obviously i didn't hack hardlink dir support into your kernel via python, so the dir structure is still duped. (wasting a few KBs of hard drive space, i'm sure, but what the heck)

no inotify, so i do have to scan the entire dir during every back, but this hasn't been prohibitively long-running even on very large dirs i've tested it on.

And rsync certainly does not facilitate multiple hardlinks to a directory to be made.

nope, but to files it does, and that's what uses the most disk space. check the option: --link-dest

Therefore, this shameless plug probably does not offer something similar to time machine. Unless the author also claims that a Trabant is something like a Ferrari.

italian trash... =P

sheet music editor != composer

[FranTaylor]

I know lots of musicians, and they create music, not pdfs. And none of them compose in sheet music. Many of them are linux users, but none of them use linux for composition. There is nothing on Linux that even comes close to Garage Band for the non-geek.

Re:sheet music editor != composer

[tomstdenis]

Ok, I have no idea what you're talking about. I write my music usually in a combination of on staff sheets and on the computer (using Lilypond). Granted, I'm not a pro composer, but I can't imagine doing it another way. You have to at some point put your music on a staff sheet. Otherwise, how can others read it?

Lilypond can also make midi renders of the music which is useful to quickly hear something back (granted lacking dynamics).

Anyways, I disproved your little notion there. I'm not only a composer, but I'm a composer who uses Linux, and not only linux, but I use Ubuntu.

Tom

Re:I assumed a kind of simplified tape backup syst

[Anonymous Coward]

I don't think you quite understand what TM is backing up. While it does indeed do a "full backup", it would not write your 44GB every time it saves, simply write those files that changed. So unless you are changing every file in your system, your 300GB disk will last much longer than a couple of weeks.

Where a real problem would come is if your hard drive is full of only a couple very large files, such as for video editing, in which case you probably need more than 1 300GB backup drive anyway.

It doesn't work like you think.

[heinzkunz]

Time machine doesn't respond to each file change. On it's hourly run, it checks a log file managed by FSEvents for changes. Think of dnotify instead of inotify. Most users will appreciate the hourly backups. It means they can at most lose 1 hour worth of data instead of 24 hours worth. During the backup, disk io will slow down for user tasks, but apart from that, there is no noticably performance drag.

Re:I assumed a kind of simplified tape backup syst

[Sentry21]

Time Machine looks like a simplified interface to something like Netbackup or (more likely) Miranda to me. It's a handy thing to have natively but I was less than happy about how it works out of the box: in short it claims to take weekly full backups plus incremental changes every ten minutes.

Time Machine keeps:

• Hourly backups for the past 24 hours
• Daily backups for the past month
• Weekly backups until your backup disk is full

When the disk is full, it deletes the least recent backup(s) as necessary to complete the most recent backup. You may optionally receive a warning when this is about to occur.

The weekly full backups are taken until the disk is full (Apple's words). I can see that a lot of people are going to get caught out by this: on my Mac Mini the full backup came to 44Gb, which on an empty 300Gb external drive (which seems to be about the most common size sold in Maplin and PCWorld these days) will come to five or six weeks of full backups.

Because OS X now supports directory hardlinks, it is possible to keep full backups while only taking the disk space for new changes. I plugged in my TM disk after about a week last night, and it backed up 1.8 G of stuff. A few minutes later, it backed up 52kb of stuff. A few minutes after that, 15 megs of stuff. Each of these directories looks, when you browse it in the FS, as though it takes up the entirety of the space required for a full backup, but in actuality, it only takes up the space required for an incremental backup. I pastebinned an example of this last night. the 'sparsebundle' file is a disk image containing the directories indicated. Note that despite having three backups of 9.1G, one of 1.8G, and one in-progress of 5.3G, the total size of the disk image is a whopping 12G.

Thus, we can see that backing up your 44G mini will take 44G initially, and then each additional backup will only take the space that a delta would require - if you change 200M of files, your next backup takes an additional 200M from the drive, not an additional 44G.

There's no indication of what it does then: does it overwrite the oldest backup? Are you prompted to delete the oldest backup? Does it suggest you go and buy another drive? In an enterprise environment decisions would have to be made about what was kept and for what duration: you wouldn't keep everything indefinitely.

As mentioned above, it deletes the eldest, optionally prompting you.

I'm going to have to see if it's possible to write some usable rules this weekend: off the top of my head, keep music and video and photos and the like backed up regularly and maybe only back the OS and applications up periodically, in the case of the OS particularly before the 10.5.1 update as I have bad feelings about it. I think there might well be some wailing and screaming from people who have just switched it on in a few weeks time when their external drive is full...

I don't see that situation happening for months, possibly years depending on their usage patterns, and the people who enable it and have it backing up their 5G torrents left and right will soon learn to add ~/Torrents to their exclude list.

BTW, worthy tip for anyone who's looking to enable Time Machine - exclude ~/Library/Caches. Seriously, it's gigs of crap for nothing, and it can all be regenerated when needed. I hardly need TM to back up my Safari browser history.

rdiff-backup?

[Anonymous Coward]

As expected, there are a lot of mentions of rsync, but rsync alone isn't going to get you the versioning feature that comes with Time Machine. I use rdiff-backup for versioned rsync.

Restoring my just-deleted file: rdiff-backup -r now host.net::/remote-dir/file local-dir/file
Restoring a file to 10 days ago: rdiff-backup -r 10D host.net::/remote-dir/file /tmp/file
Viewing changes to a directory in the past 5 days: rdiff-backup --list-changed-since 5D out-dir/subdir

And yes, I use this even on my Macbook. Time Machine works poorly with File Vault, the disk encryption feature paranoids like me use on portable computers, but rdiff-backup does just fine.

2003, actually...

[kylemonger]

http://freshmeat.net/projects/snapshot/ And I'm sure I saw a similar, but less well documented, project before this one.

FLOSS' failure pathos

[Burz]

It is largely due to the difference in pro vs. casual development methodologies. They both have their failing in terms of discipline, but the casual crowd lack something ELSE:

* They are coding to impress/please either A) themselves or B) their coder and sysadmin peers

* They rarely establish who their audience is (not consciously).

* The above determines what interface (in the broad sense) types will be honored. In the case of GNU system hackers, the interfaces are CLI and libwhatever APIs. ABIs are shunned party because of licensing, but also to a large extent because their audience is unconsciously assumed to be people familiar with linkers and compilers.

* Press Ctrl-Alt-Bksp in your Linux GUI and watch as all unsaved data is *instantly* oblitterated. GUIs are a nuisance to the Linux crowd because they represent an unwanted commitment (or the suggestion of a commitment they want to avoid). Likewise WRT a graphics subsystem that isn't properly configurable given any reasonable set of usability criteria... if Xfree and Xorg can't write a comprehensive configuration panel themselves, they should at least give others the tools to do so (and the semantically f--ked xorg.conf file is no such tool). Any fool writing for usability would have at least provided an API to have settings changed/validated in memory AND saved to disk by the graphics subsystem itself.

And just what is a computing 'interface'? It is defined as a commitment to a set of functionality for a particular class of actors (users). But FLOSS coders for the most part are NOT going to commit to end-users or other non-peers.

The ones that buck this attitude and inherited a lot of professionalism and user-focus from the proprietary world are: Mozilla, OpenOffice, MySQL among others. Ubuntu could qualify, though they aren't my ideal of 'coders'... they have internalized some specific user-focused methodologies.

Ones that are stereotypically slacking in the FLOSS manner: Gimp, Linux kernel (sorry, that is my honest assessment and I can cite Torvalds' own comments that would support it). KDE and to a bit lesser extent Gnome (although I hate to say so, because Gnome is so technologically awful).

Recommendations:

* Learn Rational Unified Process. Particularly the parts that cover documentation of requirements and use cases; continually reviewing/refreshing these is the best discipline for staying audience-focused.

* Dump the software repository mentality. Repositories/distros are inserting themselves between you and your end-users. You will have to code your stuff with LSB Desktop as a target and include all the rest that the spec doesn't supply, but you will finally be able to realistically distribute apps to end-users independently, on your own, without dependency issues.

* Re-connect with Personal Computing as a concept. Why were IBM PCs coveted in offices that were bursting with IBM mainframe terminals? Answer: Personal initiative and control for non-experts. Distro/repository culture comes from the computer science lab and server room. The concept of PC platform, a defined and easily navigable space upon which you place third-party products, came from the garage and the backoffice. Apple has a PC platform, and MS does. FLOSS has BSD and Linux distros where a novice programmer cutting her teeth has little chance of quickly and easily sharing her work with classmates and friends (even when they run Linux)... she'd have to devote weeks to fighting with and learning the package manager/dependency tarbaby first.

* For the OS people: Your graphics and sound subsystems must be refactored before they can meet basic user expectations. Linux *still* does audio-blocking on inexpensive hardware even with ALSA; the architecture of audio mixing must be reworked to never allow blocking unless an app jumps through hoops to do so. Not hearing a softphone ring, or a meeting alarm because of some Flash-laden web pages doesn't cut it... it's UNUSABLE!

Fixing this w

Re:FLOSS' failure pathos

[ealar dlanvuli]

I concur with everything you said.

Too bad the linux kernel guys will never address their ends of these issues.

Sean

Re:FLOSS' failure pathos

[dbIII]

At some point it becomes clear that reading and writing can communicate more than pointing at pictures. This also applies with computers and is why the command line is useful and why a mixed command line and GUI environment is useful for a variety of tasks. Some other bits of the rant are valid - but the limiting nature of GUI's makes them really hard to put together for flexible tasks unless it's a mixed situation (eg. google search text box which allows text entry and search commands). There's some MS Windows shareware that duplicates most of the configuration options in the diplay portions of Xorg that demonstrates how complex a GUI can get when you have a lot of options (can't recall the name sorry) - hence the cut down versions and sometimes flawed GUI config tools from each distro and just a text file from Xorg itself. You might not need six video cards in the system but some people do things like that and how do you add that all into the GUI? Having a text file that can be appended to adds flexablility.

Personally I see GUI only tools as a step backwards to the first days of the Atari ST - if you want to do things on a few dozen machines you do not want to log into each one and point at pictures for ten minutes. Also from the days of Win3.1 (or before) it has been blatantly clear that GUI only display configuration is an utterly stupid idea and that it is only advances in monitor hardware that have enabled people to get back from bad configuration choices instead of having to work out which of six mouse pointers is the real one.

Re:FLOSS' failure pathos

[Burz]

GUIs are indeed chock full of not only icons and text, but keyboard commands and CLIs as well. Text may communicate more than icons in some situations, but text + icons together are so much more useful that they've become the standard; For the latter, you need a GUI. And it has to work.

But we have a particular GUI that is unmanageable for ordinary users

You might not need six video cards in the system but some people do things like that and how do you add that all into the GUI? Having a text file that can be appended to adds flexablility. I never argued against a text file. But Xorg itself needs to be the default method of managing it, for 'dogfood' reasons at least. Maybe someday they will even be able to code a graphical configuration tool, but I think a resizable 'X', a clock and calculator are the state of their art for this epoch.

... it is only advances in monitor hardware that have enabled people to get back from bad configuration choices instead of having to work out which of six mouse pointers is the real one. The only bad configuration choice that should ever leave someone without a GUI for display reconfiguration would be connecting a monitor that doesn't support the minimum requirements of the OS. As for pointers, I have never seen that problem, though I have booted ostensibly English-localized Linux distros that (mistakenly) required a different language keyboard when switching to console (rendering the system unmaintainable in that case -- if a GUI tool had been present, then it could have worked out).

Bottom line is: There is no excuse for throwing an end-user into text-only environment for any reason whatsoever. The GUI can do it all, even if temporarily falling back to an industry-standard base mode to correct a display problem.

Re:FLOSS' failure pathos

[dbIII]

Bottom line is: There is no excuse for throwing an end-user into text-only environment for any reason whatsoever

I disagree a great deal with this especially when it comes to display configuration, servers and embedded devices. I've seen over a dozen situations where the display could not be configured correctly without pointing and hoping on windows 95 alone due to the screen being out of range. I especially do not want to have to go through the same GUI sixteen times in succession when I could just copy the same file to sixteen computers with no editing required.

This bizzare idea where we have to always point at pictures instead of reading and writing limits things especially when you want to do a lot at once. Text has a place.

Re:FLOSS' failure pathos

[dbIII]

Looks like I missed this bit:

Maybe someday they will even be able to code a graphical configuration tool, but I think a resizable 'X', a clock and calculator are the state of their art for this epoch

Facts are not going to really stop this exercise in fantasy are they? I wonder if the language thing was also somewhat beyond reality.

I'm curious - where is the industry standard base video mode on a Nintendo DS or a telephone? There is Xorg running on such things. Look beyond the glass typewriter on the desk - computers are used for a lot of stuff and pictures to point at for absolutely everything is not always desireable.

Re:FLOSS' failure pathos

[Burz]

I've seen over a dozen situations where the display could not be configured correctly without pointing and hoping on windows 95 alone due to the screen being out of range. Win95 is positively ancient and did not automatically fall back to safe video mode.

This bizzare idea where we have to always point at pictures instead of reading and writing limits things especially when you want to do a lot at once. Text has a place. I already pointed out that I never said the text config should be eliminated. This is typical anti-GUI kneejerking, assuming that a text feature will be taken away when a GUI feature is made standard.

Text has a place... on a bitmapped display.

Re:FLOSS' failure pathos

[Burz]

The context of the discussion is desktop computing. Looks like you missed that too.

I'm curious - where is the industry standard base video mode on a Nintendo DS or a telephone? Those devices don't depend on separate displays-- they are built in so relying on an industry-standard config is moot in those cases. Can the questions get any more paranoid and inane?

Re:FLOSS' failure pathos

[dbIII]

Fine - discard counter examples to the off topic discussion about open source as offtopic and insult the person giving them as paranoid. Game to Burz since I've just realised that it all it was.

Re:FLOSS' failure pathos

[Burz]

LOL!

This story has everything to do with open source and desktops. Go back to the top and re-read it.

You are paranoid. You're seeing so much red that you can't even realize there is no difference in terms of monitor support between an 80x25 character-mapped display, and a VGA bitmapped one. Yet you insist that deprecating character-mapping equals taking away text.

The real contrast here is between text-only mode vs. bitmapping where text + graphics are equally available. But you think in terms of a (false) choice between text-only and icon-only... which is classic CLI-jockey paranoia. And its so thick its robbed your ability to think straight.

Time Machine problems with FileVault, large files

[markjhood2003]

Apparently it doesn't work well with FileVault:

http://blog.nominet.org.uk/tech/2007/10/31/leopard-and-filevault-wont-work-well-with-time-machine/

I also don't see how it's going to work well with the disk images associated with virtualization technology like Parallels, or with large monolithic email folders and databases.

The problem is that Time Machine backups at the file level of granularity. Block-based backup technology like Ghost is a much better solution -- it can run in the background, perform full and incremental backups whenever you want, and the backup sets can be mounted as drives for browsing. And your backup disk won't fill up with multiple copies of your email or VM disk images because you forgot to exclude them from your backup.

Time Machine is Kewl

[qazwart]

I got Leopard just a week ago, and Time Machine has already saved my butt. Before Leopard, I had created a cronjob to to a compressed tar of my entire system, and put it on a secondary hard drive. I backed up three times per week with a rotating backup. Restoring was pretty much untaring that one file I needed. I could do it, but my wife certainly can't.

I've seen software on a NetApps that is "similar" to Time Machine. In your $HOME directory, there is a ".snapshot" directory that contains 24 directories (one for each of the past 24 hours) and a month's worth of days. If I needed a file I deleted 2 weeks ago, I could go to the correct "day" directory, and see what my $HOME directory looked like at that time. I could search the .snapshot directories using the standard Unix tools too, so if I don't remember the version of a file I needed, I could do a find and a grep to find the correct file. However, this still wasn't Time Machine.

If I was looking for a missing file on my .snapshot directory, the "find" command would take forever because I was combing through all of those directories one file at a time. You type in a missing file name in Time Machine, and in seconds, that file is found. Plus, Time Machine doesn't have to work on a "file" level. As someone pointed out, I can search through my address book for a particular set of records, and only restore those records.

Plus, I am sure that the technicians who setup the NetApps took a lot of time and effort to configure the ".snapshot" directory. Time Machine took me about 30 seconds to setup. (Step #1: Select backup drive. Step #2: Click the "On" switch. Step #3... There is no Step #3).

Time Machine is and isn't a marvelous technical breakthrough. There's nothing about Time Machine that is earth shattering technologically. However, the way it is put together is amazing.

I am sure that we will see many copycats in the next few weeks. I am sure that Microsoft will come up with something very similar in a few months. After all, the backend of Time Machine is quite simple. It's the front end that's absolutely amazing.

Here's one for Solaris

[BrainInAJar]

Time machine for Solaris: # zfs snapshot mypool/myfs@`date +%m%d%H` ta da... no hackery involved, it's a shipping feature.

Error in code

[dotancohen]

Redhat/Fedora: $ yum install pygtk2 gnome-python2-gconf gnome-python2-extras pygtk2-libglade That should be:
$ su -
# yum install pygtk2 gnome-python2-gconf gnome-python2-extras pygtk2-libglade

Note the hash before the command, as opposed to the dollarsign. The hash indicates that we are root.

Re:Error in code

[dbIII]

It doesn't matter! It is a prompt! It is not "code". It can be redefined to be different even if the default might be a hash in fedora with the bash shell.

OK - now assuming it is bash (it doesn't have to be to call yum like this), say root has the following in their resource file:

export PS1="$"

Now it is a dollar sign.

export PS1="\[\033[32m\]${USER}\[\033[0m\]@\[\033[31m\]\h\[\033[0m\]:\W>"

Now it has colour.

Re:Error in code

[dotancohen]

And apt-get can be an alias for yum. So what? You can customize it all you want, but then it's no longer Fedora. The site had instructions for Fedora and there was a minor, minor, minor error that I caught. I'm certain that 50 thousand other /.ers caught the error as well, decided that it was so minor that it was not even worth mentioning, and went on with their lives.

Re:"something like"=/=real thing. technology missi

[Game_Ender]

Why don't you use http://pyinotify.sourceforge.net/ to enable inotify support?

She's right you know

[Anonymous Coward]

And so she should wash your mouth out with soap. For the crime of a stupid date format! You'll be sure to use 2007-01-11 (assuming you actually meant January since Americans get even this part wrong) when you write this code, right? 11-01-07 could mean anything, including the 7th of January 2011. ISO8601 forever!

Re:She's right you know

[snoogans126]

Um, I'm a bit confused by your saying that Americans get this part wrong.
If you start with the position that ISO8601 yyyy-mm-dd is the "chosen one" of date formats (which I personally do), the common American form m/d/yy or m/d/yyyy, merely has the year out of place.
The common "everyone else" form of d/m/yy or d/m/yyyy is far more out of wack.
I know discussing date formats can be as productive, as discussing Jesus vs. Budda vs. The Flying Spaghetti Monster. But I don't think that the common American date form is any less wrong than anything else that isn't ISO8601.

How flexible is it?

[el_chupanegre]

I was wondering about upgrading to Leopard once I get a new external HDD, but I have one big reservation, and I want to know if anyone knows the answer to this.

Does the Time Machine drive _have_ to be HFS+? I want to use Time Machine to manage my backups for it's ease of use, but I want to be able to also treat my external HDD as an external HDD! I want to be able to take it to someone's house who doesnt have a Mac and access my files from their machine (probably Windows XP obv).

Like most of us here I'm the local geek who fixes everyone's PC's and I stick alot of useful software on a DVD at the moment for helping fix someone's computer. I'd prefer to have this on an external drive (as well as all my backups) and use it.

If it insists on HFS+ it looks like I'll be using BackupPC. A nice interface is great, but if it makes my external HDD fixed to this machine, what's the point?

I still haven't thought what to format it anyway if I don't go for Time Machine. Obviously HFS+ can't be used everywhere, NTFS is no good on my Mac, FAT32 can't be used for backups with the 4GB file limit. What am I to do?!?

Re:How flexible is it?

[Anonymous Coward]

So partition it. Create an HFS+ partition for time machine (yes, it's required--duh!) and another partition for your other data (NTFS, FAT32, ext3).

FreeBSD has "Snapshots" too

[RevDigger]

FreeBSD has had support for "snapshots" in the base install for some time now. I believe it was modeled on a feature in NetApps. Other file systems may support a similar feature. Basically, you tell the system to take a snapshot - say every hour via cron - and it starts a new log of changes that have taken place on the file system. Multiple snapshots may be maintained concurrently. With a little snapshot magic you can mount, and even automount these just like NetApp and Leopard. It works great. Note that it saves the snapshot on the same filesystem, so it does not get you out of making proper backups.

The beauty of the OSX's time machine is that it works right, by default, without any setup (If you have an external drive for the purpose). The upshot is that in 5 years, maybe 50% of mac desktops will have SOME kind of regular backup. That will be a pretty huge jump from about 0% now.

Re:"something like"=/=real thing. technology missi

[slashflood]

Sorry, but please check out BackupPC. All the features you named are part of BackupPC since its existence. I know, the concept of doing Backups by linking files to the backup space is something new to a Mac user, but it doesn't mean that Apple is the inventor. BackupPC does more than this, but I don't want to get into thos pesky details like compression and pooling (very nice invention!). And of course you can access the backup pool with your standard utilities. You say, rsync is not TM and you are right, but BackupPC is much more than rsync.

Careful with your linguistics there, cobber ...

[shellbeach]

...Also, TM is not confined to the Finder per say. if you're in Address Book and lost a contact, type in the filter string to locate it. Still... It's PER SE, goddamnit! And it means "intrinsically!" You saying "TM is not confined to the Finder per se" would imply that either it IS somehow confined to finder (but not intrinsically) or you just like to use big-person words you don't understand. Hmmm ... I think the interpretation would be, "TM is not confined to the Finder by itself, but also relates to the Address Book and other applications." The use in this sense is actually most correct - the literal translation is "by itself", and this has been adapted in English to mean "in and of itself" or "intrinsically". Nevertheless, dictionaries still list the definition as "by, of, for, or in itself; intrinsically" (e.g. http://dictionary.reference.com/search?q=per%20se, and see http://www.etymonline.com/index.php?search=per+se&searchmode=none).

If modern use of "per se" decides to move back to "by itself", that's fine by me. But even if society decides to start using "per se" to mean "fucked if I know" - and uses it consistently in this fashion - there would be nothing inherently wrong with this. Language evolves and is fluid, and we everyday use hundreds of words whose current meanings have no relation to their original etymology. We even use words that now mean the exact opposite of their roots - such as "philanderer", for example (see http://dictionary.reference.com/search?q=philanderer).

Language is a means of communication. If people understand what you're saying, the rest is simply semantics ... :)

Ext3cow

[flithm]

First of all this isn't really comparable to time machine, as it's just another rsync based backup system. That being said there's ext3cow that's been available for Linux for some time.

It's a work in progress, but it's already quite stable. Maybe if more people get into it, development speed will increase. There's even a simple GUI available.

Re:Ext3cow

[wolf08]

Interesting that the creator of that simple gui now works for Microsoft on their kernel team.

Using my highly accurate* skills at seeing the future, I have determined that Microsoft is not only going to have a kernel for windows 7 (vista+1) that uses the concept of ext3cow, but also will be written in Hindi, and will be completely open source.

Ok, being a bit more serious... I bet that since this guy (who is very interested in data preservation) is on the windows kernel team, and has been since 2005, will influence data storage in future versions of windows.

And as a footnote, I'm sitting here on my Kubuntu 7.10 installation, being bored and checking out backuppc (which looks fantastic!)

*highly accurate in this context is defined as completely speculative

Mercurial :-P

[gcantallopsr]

cd
hg init
hg add
echo '@hourly hg commit' | crontab

Ooooh! Time Machine! X'-D

Just add a few more lines and parameters, and it's actually usable :-P

different market

[sentientbrendan]

There's obviously lots of version control systems for developers (svn) and business people (windows shadow copy) and there's even version control built into some applications like word, although few people know about it. The point of time machine is that it is rough grained general purpose version control for the average user who can't deal with the complexity and rough edges of svn. Version control for movies, pictures, and word documents with a slick interface.

This is really what apple is all about, taking existing technologies and making them accessible to joe user and useful for the sort of problems that joe user addresses. This simply reflects that apple is pretty much entirely a consumer technology company, whereas linux and even windows to some extent tend to be oriented towards professionals.

People tend to learn about windows at work, and then end up buying a windows home system, whereas people tend to learn about macs at home and, sometimes, end up bringing them to work.

give it a few months

[m2943]

I'd be surprised if you didn't see a Time Machine-like backup system for Gnome within the next few months. I also suspect that the Gnome developers will learn from the Time Machine implementation may be able to come up with a better underlying implementation.

rsync

[barl0w]

Very cool. Great work.

I've played with Leopard for a few days now, and is it me, or is it a close knockoff to Solaris/Any favorite Linux distribution? I'd almost rather use Ubuntu than Leopard. I mean spaces has been around forever, rsync instead of Time Machine, and I just hate browsing in the Applications dir for the programs I haven't docked yet. I wish there was a Win key + E alternative to getting to browsing the files system, like there is in XP too.

I'm not against X, as a matter of fact, that's the only laptop or desktop that I'd put money towards right now for a family computer. But I find myself more productive in XP/Vista/Ubuntu after trying Leopard.

ext3cow versioning FS

[reedk]

A less GUI-driven but probably more robust implementation of a similar idea is ext3cow.
If you want the GUI for it, add on The Time Traveling File Manager.

Is Time Machine really a good thing?

[sjdude]

In this day an age of warrantless searches and National Security Letters, is it really a Good Thing to have everything that ever passes through your computer be stored forever? Or not be easily and permanently deleted?

Time Machine For Windows?

[Anonymous Coward]

Its XP only at the moment but so is 90% of the world.

http://www.rebit.com/

Same kind of "history"/CDP as time machine AND full bare metal recovery to a specific date (like if you have to replace your hard disk). No configuration/settings at all, never fills up (if your lap top disk is smaller the the Rebit).

Maybe Apple is the one playing catch up.....

Plan9

[mattr]

I once read that Plan9 used a WORM drive and basically all work from the office was saved at byte-level granularity. This is really what I want.

A Time Machine like system would be useful, but there are two problems I see. One is that a file could be corrupted if being written to when the backup is made (at least someone said this happens). FWIW there's something called pdumpfs from 2004 like flyback (in Ruby).

The other is of course that my most common needs for backup are erasing something by accident, "organizing" my files and then deleting something to where they've been organized, somehow losing my current version and wanting to revert to the version a minute before, etc. Other uses are when I have files on old media I need (where's the memory stick reader and the memory stick, or what's in that dead computer), and also of course the need for remote secure peace of mind backup (not on an ISP).

Time Machine doesn't answer all these things. I think there needs to be a system-wide stream backup daemon that apps can call, then the apps need to support it. Definitely I would like OpenOffice to do something like this that would work cross-platform. a series of graduated scales allowing one to zoom back quickly would also be good, finest grain one can show context. Apps should also provide hooks to the OS so that such an OS-supported interface could for example compare layers or bitmaps in the Gimp to show what has changed, compare stroked vector art, compare rich text, etc. in an intuitive manner perhaps using outlines or coloration so you can grab bits of what has changed.

All that said I think I would be very happy if 1) OpenOffice saved the entire keystroke/mouse event stream allowing reversion at any granularity, and 2) common filesystem commands could be relinked to utilities that maintain a trash space so you can revert prior to your (my) latest fuckups within the past week, say. These things do not seem too difficult technically.

Saves much disk space

[gottabeme]

A huge benefit of rdiff-backup is that after the first backup, it only stores changes to files. Someone once gave the example of backing up a 2 GB Entourage mail store with Time Machine. Each time that the mail store changed, Time Machine would have to add 2 more GB of data to the backup disk. rdiff-backup would only backup the changed data in the mail store, saving huge amounts of disk space. It would take a very long time for the diffs to add up to the size of the mail store itself (unless you got a WHOLE lot of large messages real quick), and would likely take much less time to run the backups too.

Preview

[Per Abrahamsen]

Again, I have usually seen retrieved files (and directories) being placed in a different location than the original, so you can examine them before manually placing them in their original location.

Re:Preview

[MightyYar]

Right, but compare the workflow -

1. You're working in iPhoto/Picasa and you realize that you are missing some pictures.
   
2. Click on time machine and browse through to find the last time you see the pictures.
   
3. Select the missing album/pictures.
   
4. Click out of time machine and continue working, with your pictures magically restored.
   

-OR-

1. You're working in iPhoto/Picasa and you realize that you are missing some pictures.
   
2. Track down the directory that iPhoto/Picasa uses to store its pictures.
   
3. Create a temporary directory to restore to.
   
4. Right click on the Pictures directory and select properties and versions.
   
5. Select a version that you think might contain the photos.
   
6. Restore that version to the temporary folder
   
7. Look for the pictures.
   
8. Repeat steps 4-7 until pictures are found.
   
9. Import missing pictures back into iPhoto/Picasa
   
10. Fill back in meta data that was lost
    
11. Clean up temporary folder.
    

Even for applications that do not support the feature, it saves time and effort by allowing previews and not creating any temporary files or folders that you then have to clean up (or if you are like me, forget to clean up). And of course Time Machine is easier to set up for the typical user, who is just going to buy a big honking external drive and plug it in.

What BackupPC does...

[SanityInAnarchy]

I don't know about Time Machine, but I have set up BackupPC for work.

Here's what it does that's unique:

Incremental and full backups with rsync. (Too bad it forces you to do "full" every now and then, a waste of bandwidth, but read on...)

Compression and pooling. Every file backed up is first bzip2'd, then hashed, before it's stored. By checking that hash, it will only store each copy of a file once. So, this handles if you have the same file three times in a single backup (ie three separate locations in your home dir), or if you delete a file and then restore it later, either by accident or from somewhere else.

I imagine Time Machine could be like that, though I'm not sure. If it relies on ZFS, it would at least be faster.

To be fair, what BackupPC cannot do is store incremental backups on the same machine efficiently. You can do an rsync backup with hardlinks, and as long as you can be sure that anything overwriting a file will delete it or rename a tempfile on top of it, you can basically have anything that hasn't changed be stored only once.

But, BackupPC and native rsync are both slow with lots of files. It has to be a LOT of files, but the issue is still there. I've actually run out of RAM when trying to backup a BackupPC repository itself with rsync, which is why it now uses drbd instead.

What's really needed -- and what Time Machine _may_ provide, though again, I don't know, without picking up a copy of Leopard -- is support for this kind of thing at the filesystem level. Basically an auto-CVS (or SVN, at least). I'd be happy using a distributed VCS locally -- something like bzr or mercurial, though it'd be nice if it could be as smart as BackupPC about multiple copies of the same file -- but that still means moving (renaming), deleting, creating, etc all have to be done manually (bzr mv, bzr cp, bzr rm, etc). Some FUSE module would be great...

There's still one thing lacking from that -- cp is still inefficient. Would be really nice to have it copy-on-write. But, I guess that's not happening without changing POSIX, which I guess isn't happening.

How does that work?

[SanityInAnarchy]

I'll be willing to bet that it's not a "hard link" in the strictest sense. You see...

The problem with hardlinks to directories is, you can create loops that way.

If you allow this, you have to use a garbage collector for your filesystem, instead of (or in addition to) a simple reference-counting scheme. Or you can just leak disk space...

A naive garbage collector would basically involve searching the entire tree for orphaned folder hierarchies, and freeing them.

And not allowing this is tricky.

For most cases of needing a hardlinked directory, Linux can simply use symlinks. For cases where symlinks won't work (because some program is actually checking for them), a bind mount will work.

But I don't think either of them cover the intent here, which is to make a copy-on-write image of a directory tree, not a hardlink. After all, if it was a hardlink to the current state of the rest of the FS, and I rename a file in my home directory, it'll be renamed in the most recent backup, too. Maybe it's only between backup revisions?

It's all about where you draw the line.

[SanityInAnarchy]

For example: Was Quake a genius program, or a ripoff of Doom?

If by "design part", you mean strictly graphics, you're clearly wrong about Time Machine.

If by "design part", you mean the high-level design of the technical underpinnings -- for instance, the way in which Time Machine watches the filesystem for changes, and makes all revisions available as part of the filesystem at every level (so that only Time Machine can tell it's not actually making a full backup of your entire hard drive once an hour)...

If that's your definition of "design", you mean software design, which is quite a bit different than UI design. When most people say "design", they generally mean that UI part, the fancy graphics and so on.

And if that's your definition, you're still wrong, because the "Linux version" is actually more about the UI, and is actually technically less advanced, even if they had the same interface. So no one "made it for Linux".

Branching? Copying?

[SanityInAnarchy]

Efficient copies are the hardest, as POSIX does not -- in fact, I don't know of ANY filesystem API that lets you create a copy-on-write file.

Also, how are you planning to handle branching?

I realize you're talking a light tool, but I'd like light and uncrippled, if possible. Because if you add these things back in, you would basically have everything you need from svn -- or a local one like bzr.

(If you're vigilent enough to use CVS or SVN, by the way, just use bzr for local-only stuff.)

Mod parent up

[gravis777]

That is the most useful and insightful description of Time Machine I have ever seen

inotify != fsevents

[LKM]

inotify and fsevents are different, as I understand it. inotify calls applications if something in the file system changes. fsevents doesn't do that; instead, the application calls fsevents, and fsevents returns a list of files changed since the last call. For something like backup, the second solution is obviously better, since it creates less overhead.

So there's no performance drag in OS X, but there would be one in Linux if a similar solution used inotify.

Re:inotify != fsevents

[hummassa]

fsevents is a daemon that goes polling the OSX equiv of inotify and populating a log. So, where exactly is the overhead??

Re:inotify != fsevents

[LKM]

Not sure I understand your question. I intended to say that using fsevents creates less overhead than being called by the OS X equivalent of inotify directly. The overhead in being called by inotify directly is that your app needs to be called every time the FS changes. If that occurs often and your app isn't extremely efficient, this may drag the whole system down. Using fsevents, only fsevents gets called when the FS changes. Your app can then get a list of changed files from fsevents whenever it feels like it, and doesn't need to worry about efficiency.

Re:inotify != fsevents

[hummassa]

For starters, let me say that I had no intention of rattling anyone's cage ;-) It was an honest question.

Not sure I understand your question. I intended to say that using fsevents creates less overhead than being called by the OS X equivalent of inotify directly. The overhead in being called by inotify directly is that your app needs to be called every time the FS changes. If that occurs often and your app isn't extremely efficient, this may drag the whole system down. Using fsevents, only fsevents gets called when the FS changes. Your app can then get a list of changed files from fsevents whenever it feels like it, and doesn't need to worry about efficiency. First of call, events in /dev/fsevents (OSX equivalent to inotify) go quite often.

my math:
case 1: total OH = OH of going to /dev/fsevents from time to time, in your program
case 2: total OH = OH of calling fseventsd + OH of fseventsd going to /dev/fsevents from time to time

in which case you think the total overhead is greater?

Re:inotify != fsevents

[LKM]

case 1: total OH = OH of going to /dev/fsevents from time to time, in your program
case 2: total OH = OH of calling fseventsd + OH of fseventsd going to /dev/fsevents from time to time

in which case you think the total overhead is greater? Ah, okay, I see the issue. Here's the original quote which triggered my reply:

Actually, I have mixed feelings about having a daemon following inotify (fsevents equivalent for linux) in order to backup. What I meant to say was that Time Machine use the FSEvents API instead of reading from /dev/fsevents. Time Machine does not do the Mac OS X equivalent of following inotify. There's no event notification involved in Time Machine. Instead of being notified of each change when the change occurs, Time Machine calls the FSEvents API once an hour (I guess) and gets back a list of changes that need to be backed up.

I wrongly wrote "instead, the application calls fsevents." I meant to say that it uses the FSEvents API.

Since fsevents notifies all clients each time the FS changes (as far as I understand it - I don't program for Mac OS X), using it is quite often not a good idea. Furthermore, fseventsd does not write all /dev/fsevents events to its log; it only logs changes on the directory level. Thus, I would say that your case 2 is more efficient.

File notification

[jesup]

One of the first systems (probably not the first) to make use of filesystem notifications was the Amiga; I added Notification support to the Amiga ramdisk Way Back When for OS 2.0 (3.0?) so that applications could ask for notification on changes to system or application preferences stored in the virtual PREFS: disk (normally stored in RAM: after boot).

The equivalent would be having something in X (perhaps in KDE/Gnome/etc on top of X) watch the xorg.conf file, and when it changed re-read and re-parse it. So "cp xorg_1024x768.conf xorg.conf" could cause an automatic resolution change to a saved 1024x768 configuration (assuming X handled on-the-fly resolution changes) - or that if you wrote a new xorg.conf editing tool, all it had to do was rewrite xorg.conf to get the changes to take effect. And other programs could be watching it as well and react, instead of having to somehow know that programs X Y and Z are interested and need to be HUPed or stopped and restarted or whatever.

All the daemon "HUP-to-reread" stuff would become automatic with this sort of setup.

Ancient history, though many of the abilities mentioned don't exist in many (most) modern OS's. In some cases, more complex and more capable mechanisms have surpassed the old Amiga ideas, such as systems like dbus - but the existing Linux software doesn't always make good (or consistent) use of them!

Comment removed

[account_deleted]

Comment removed based on user account deletion