Ask Slashdot: It's World Backup Day; How Do You Back Up?

MrSeb writes "Today is World Backup Day! The premise is that you back up your computers on March 31, so that you're not an April Fool if your hard drive crashes tomorrow. How do Slashdot users back up? RAID? Multiple RAIDs? If you're in LA, on a fault line, do you keep a redundant copy of your data in another geographic region?"

Comments

Time Machine

[anethema]

Apple hate aside, time machine is an amazingly excellent backup system.

It backs up to a Netgear Readynas configured in RAID 5. Hourly, daily, weekly backups. I've never lost anything thanks to this great system.

In linux I try to approximate this with BackupPC.

http://backuppc.sourceforge.net/

It is really an excellent piece of software, though no where near as refined of course. You pretty much only get daily backups though since the kernel in linux does not track filesystem changes so hourly backups would be very prohibitive.

Re:Time Machine

[Ash-Fox]

You pretty much only get daily backups though since the kernel in linux does not track filesystem changes so hourly backups would be very prohibitive.

inotify(7)

Re:Time Machine

[jo_ham]

I'll second this. I use Time Machine too. I don't have any fancy NAS box for it (due to budget mostly) - I just use an external firewire disk right now, and it has been used once due to a full internal drive failure (restoring the iMac back to the state it was in an hour before the failure) as well as the occasional single file recovery.

Most back up systems work well for full system recovery - Time Machine is not unique there - but it's the single file recovery tool that really makes it shine. It's very simple and intuitive to use.

It is totally "hands off" though - you have to trust that it actually is doing what you tell it to, beyond the menu item that gives you a summary of what it's up to (total being backed up at that moment, last backup time etc). It doesn't have a "show me a list of files backed up at x time" feature without the use of third party tools, so people who really want peace of mind may find that annoying.

Re:Time Machine

[digitallife]

This.
I have (and still do) use all sorts of various systems for backups both at home and at work, and Time Machine is by far the best. Completely invisible, automatic and smart. You can turn off your computer mid-backup and it just continues when you turn it back on. It is so much better than the alternatives, I'm surprised how little limelight it gets.

Perhaps just as important as the backups: it has a great UI to access said backups! One click gives access to a file at any date in the past you want.

Re:Time Machine

[japa]

Back In Time is a simple backup tool for Linux inspired from “flyback project” and “TimeVault”. The backup is done by taking snapshots of a specified set of directories.

I use it with external USB drive and it has saved my butt couple of times. Cases where I thought the focus is in certain nautilus window, then doing Shift-delete + enter in very quick fashion and fraction of a second later realizing there was another nautilus window with focus on some directory which is now nuked... As this is just a frontend to rsync and uses hard links, there is the advantage of the backed up files being available even without the backup program as normal files within the directory structure on the backup media.

The Tao of Backup

[Anonymous Coward]

Killthre... I mean The Tao of Backup

rsnapshot

[Wagoo]

rsnapshot seems to work pretty well for incremental rsync'd backups for me. It uses symlinks to maintain the older snapshots, to save on total filesystem usage. It can do rsync over ssh for backing up remote servers/pushing local vital data to a safe remote location.

Local backup server uses Linux software RAID for good measure (5x1TB RAID 5 + 10x2TB RAID 6).

Don't forget restore, is just as important.

[goombah99]

Backup is only half the problem. Restore is the other half. And indeed that's where I've usually had the most problems. The third problem is validating the restore. You always worry that you are either going to overwrite something on the restore target or miss something on the restore source and end up in an inconsistent state.

Time machine is revolutionary because it is so simple and seems to be almost flawless. I've had lots of backup systems over the years including dump 0 but everyone has been plagued with issues that arose when things were off normal. I've cobbled all sorts of things like rsync and cpio but the only thing that comes close to working as flawlessly as time machine is a NetApp.

At work where I can control the remote servers securley on a closed network I am able to use time machine for a remote backup. But at home I don't have a remote server I can target for the remote backup.

TO do a remote bakcup at home I use Crashplan. I looked a lot of competitors like Mosy but settled on crashplan for two killer reasons. The giant problem with all these commercial backups is that while the incremental backups are simple over the net, the restore of a whole hard disk cannot be done over the net. You have to pay them to burn DVDs and send them to you. ANd that assumes you know what time period you want to recover.

  UNlike all the other methods crashplan lets you pick a buddy who runs crash plan and then you can back up your disks to each others computer. If you need to to a massive restore you just drive over to your buddy's house and pick up the drive, bring it home, and restore locally. This also solves the problem of the first dump being too large to send over the net as well. You do it locally then drop the drive off to your buddy.

Brilliant!! plus with crash plan you pay for the app once not monthly.

I've used it for years now and it works very well and it very easy to set up. All your files are encrypted so buddies can't read each other's drives.

The only flaw with crashplan is that it runs in java so you have this instance of java running 24/7 and not to put to fine a point on it: java sucks. I don't know if it is crashplan or other things that run in the JAVA VM but over the week it bloats up to 600MB to 800MB. THe workaround solution is to kill the java VM every few days. Empirically crashplan is robust enough to survive this and restart. But that's a really awful solution.

Re:ZFS

[zippthorne]

Indeed it is not as efficient as it could be. However, using it is only slightly more complicated than "buy a usb hard drive and plug into computer"

An efficient, totally ideal process that no one actually bothers to use because it's either too complicated, or because it isn't actually licensed for your platform or whatever, is no backup system at all.

Also, ZFS is a filesystem that can be set up to preserve version information. It's not a backup while it's on the same disk....

Re:shell script

[nightfire-unique]

Just a suggestion; you shouldn't delete any backups prior to writing (and possibly verifying) your new backup. Imagine what would happen if your disks failed during your backup. It's more likely than you think; it's a period of intense I/O. I've personally had it happen during raid reconstruction.

You might consider timestamping your backups, and deleting all but the most recent 3 after a successful backup.

Something like:

/bin/ls -1tr "$drive/*.tgz.aes" | head -n -3 | xargs rm "{}" \;

.. should clean up them nicely.