Slashdot Mirror
Why Cloud Storage Is Lousy For Enterprises (and Individuals)
storagedude points to this article at Enterprise Storage Forum which argues that cloud-based storage options have fatal limitations for both businesses and individuals: "The article makes the argument that high volumes of data and bandwidth limitations make external cloud storage all but useless for enterprises because it could take months to restore the data in a disaster. It also appears to be a consumer problem — the author spent three months replicating 1TB of home data via cable modem to an online backup service." Seems like those off-site incremental storage firms could dispatch a station wagon full of tapes, for enough money. Update: Here's another reason, for Sidekick users: reader 1ini was one of several to point out an alert from T-Mobile that "...personal information stored on your device — such as contacts, calendar entries, to-do lists or photos — that is no longer on your Sidekick almost certainly has been lost as a result of a server failure at Microsoft/Danger."
Get high and enjoy the cloud !!
Seems like those off-site incremental storage firms could dispatch a station wagon full of tapes, for enough money.
Some of them do, for exactly that reason. MozyPro, for one.
its too cheap and easy to keep my files locally (more dependable too)
usb thumbdrives, CDr & DVDr even harddrives are large and cheap (both external & external)
i see cloud computing as someone with a bunch of servers owned by somebody that has run out of ideas for making money, and/or with a nose for snooping in to other people's data (i bet the government likes that - the snooping part)
Politics is Treachery, Religion is Brainwashing
I just bought a terabyte drive for $100 to back up the other terabyte drive I bought several months ago for $160. Now everything is backed up in multiple. And I can access it without getting online. And I don't have to worry about my cloud storage company going out of business and taking all my data with it.
RS
Shoes for Industry. Shoes for the Dead.
http://aws.amazon.com/importexport/
A little pricey, but handles the "station wagon full of tapes" issue.
...because it could take months to restore the data in a disaster.
It takes months to restore data from a box of harddrives? Sounds like a problem with the backup policy, not the technology.
It also appears to be a consumer problem -- the author spent three months replicating 1TB of home data via cable modem to an online backup service.
There's nothing technological preventing this from happening faster. Bandwidth limitations are artificial -- Comcast and most other cable service providers could easily provide fifty times more bandwidth to their customers, but they won't, because they're afraid you'll also use it for streaming HDTV and tell them to go stuff it with their ad-laden broadcast offerings.
#fuckbeta #iamslashdot #dicemustdie
Surely the 100$ the author "saved" by doing that could not have been worth the three months it took? That's about 140 kbps... You could buy yourself a 100$ TB drive and have a local system set to back up and restore your data whenever you need and it won't take 3 months for the data to get there and back. *And* you have control over your data and its security. *And* it would probably be cheaper anyway in the end.
Sigs are too short to say anything truly profound so read the above post instead.
One part of the cloud can go down and affect many other services that rely on it. Unless we can freely interchange services - which conflicts with the notion of proprietary services in general - cloud computing will only work when the stars are aligned. Feature sets can change, and services that depend on these will need to be updated to reflect that, but it's hard to cooperate even within the same company, much less within such a chaotic system.
Much like blogs that link to each other (or URL forwarding services for that matter), if one part of the chain goes down, you can't get to what you need. You just can't base a reliable business on that type of architecture.
Twinstiq, game news
Is this necessarily a fundamental problem or just an artifact of current systems? Seems like in the short term, this is correct, but in the long term, this sort of thing will disappear.
As a Comcast subscriber I really don't see that as their reasoning. A 720P tv show with ac3 audio and h264 video is around 1.1gb for a 45minute show. Thats around 430kB/s, or 3.3megabits per second.
In my area, Comcast has 3 tiers:
Economy: 1/0.3 for $24/mo
Performance: 15/3 for $42/mo
Blast: 20/4 for $52/mo
Ultra: 30/7 for $62/mo
Extreme: 50/10 for $100/mo
You can say they're overpriced. You can say they should offer more. But you really cant make the argument that they're preventing HDTV streaming. Anything above the Economy package has WAY more than enough bandwidth.
For a while now they've had their AWS Import/Export service. It's still in beta and only available to people in the US, but it won't stay that way forever.
http://aws.amazon.com/importexport/
Need to transfer 1TB of data? Mail Amazon the data on a drive, they load it, send you the device back. Sure beats uploading for 3 month with a cable modem. Have more data than that? You can send them up to an 8U drive enclosure, and more than that if you make special arrangements.
Boeing and Airbus are the worlds largest suppliers of cloud computing and have proven to be very reliable. Crashes are infrequent and while they can be disasterous for those directly involved they are a very small fraction of all customers. Generally replacements are on line the next day.
Haven't people rented server space for over a decade now? As for backup... One at home, one on your briefcase, and one in a remote location makes sense to me. If you trust the rented server... use it, but don't count on it.
Granted, if one has in production a data store of 1PB, and is relying on cloud storage as the backup medium, a restore of that 1PB of data will take a frightfully long time in a DR scenario. Not that there aren't many, many shops with that much data (and more) in use every day, but I'd suggest that they are the exception. I know we are. We deal with less than a TB in live production data, at most. Much of that we could live without while it is restored, because our architecture is designed with that filer-Internet-cloud network bottleneck in mind. The point is you don't have to (and shouldn't) treat cloud storage like a tape drive or a hard drive. It is something else, with certain advantages and disadvantages. To make a blanket "it sucks" statement is more than overly simplifying the issue.
I think the author is making a technical case about what is really a business matter. Most enterprises have legislative or regulatory requirements that prevent them from using the cloud. This is true here in the US (HIPPA, SarbOx, PCI-DSS, etc.), and even more true in Europe where it is a criminal offense to store certain types of data out-of-country. Companies simply must know where their data is and enforce strict controls upon it.
The article makes a few points that range from obvious to really obvious. First, backups are good. Second, offsite backups are good. The cloud isn't a big player in either from an enterprise perspective. In a traditional or legacy mode, a company backs up to tape and ships data offsite. Alternatively, some companies are using deduplication and WAN replication to move data offsite from one facility to either another facility of their own, or a 3rd party location hosting equivilent deduplication storage. The pricing and performance in the Cloud stop making sense in data measured in Gigabytes. Enterprises are surpassing hundreds of Terabytes, and moving deep into multiple Petabytes of storage. Cloud and Enterprise are at opposite ends of the spectrum when it comes to storage on a security, capacity, and cost perspective.
I see a benefit to off-site storage, and the easiest way to do that is electronically.
The second easiest way is to sync to an external hard drive and stash that in a safe deposit box or other off-site location. This can even be faster and cheaper for large data sets before the artificial restrictions on last-mile bandwidth disappear, and it avoids the problem of a backup provider going out of business.
In my area, Comcast has 3 tiers: 1. Economy: 1/0.3 for $24/mo 2. Performance: 15/3 for $42/mo 3. Blast: 20/4 for $52/mo 4. Ultra: 30/7 for $62/mo 5. Extreme: 50/10 for $100/mo
Well in my area, people know how to count.
Professionally managed redundant storage in a secure facility is more reliable than the typical home setup.
It is NOT a good idea to keep your backups with your original data. The chance of the disaster (fire, theft, water damage, etc.) that takes your data also taking your backups is greatly mitigated with offsite storage.
Hard drives fail if you are only using one drive for backup there is a not insignificant chance that some part of your backup will be damaged or have bit rot. The only way to know for sure is to restore. The only safe way to do that is with an identical machine. Otherwise you will discover your backup is bad when you test, by restoring and destroying your original.
The other problem is upload bandwidth. Surprise it takes a long time to upload a terabyte! Some simple math could have saved a lot of time. No sane person wants to have their internet connection used 100% for weeks to get that first backup.
Most current online backup companies (Carbonite, Mozy, etc.) only backup a small subset of "Important" files. This isn't a REAL backup. It is a copy of some files or as they call it a backup of important files.
What we need is the ability to do full backups quickly. We need a way to backup data without sending the data.
I think this company solved the problem and could be the Google of online backup. www.hybir.com
You don't have to choose one or the other. I don't understand why so many presumably smart people here (well, ok...) pick on a problem of some backup method or other and then conclude that it is therefore not a choice. If you really care, you have multiple backup methods - not just multiple copies, but multiple methods. They then compensate for each other's weaknesses.
Well, security issues can be another matter, as having multiple methods doesn't help your security if one of them "leaks". But I'm talking about just being able to recover the data.
I use about 4 different backup methods - some regularly and some occasionally. Apple's Time machine is real handy and I have it on all the time. That's one local copy. Mozy Pro gives me something remote in case the house burns down or whatever. It also auto-runs regularly. If I'm about to do something with extra issues such as an OS upgrade, I first make sure I have a fresh full clone using SuperDuper. And files that I particularly care about I tend to have copied onto multiple machines. If any of those methods goes belly up for some reason, I've got the others. It takes three major failures (ok, only 2 if one of them is my house going) in quick sequence to loose anything - more to loose critical stuff.
For my mother-in-law, I have her set up with Mozy (free version works because she doesn't have over 2 GB of stuff that needs backup). That's because it will happen without her attention, which is really, really important. And it also happens without me having to remember to take care of it for her regularly. She doesn't have computer stuff critical enough to need much more. If Mozy goes, I'll set her up with something different. If her computer dies right around the same time as Mozy does, then she'd loose stuff, but she'd get over it.
If the data is processed and lives in the cloud then bandwidth is no longer a major issue. As an example:
In one world you could have the Exchange servers backups pushed out to a cloud provider. This would result in many hours to get the data out there, and the challenge of restoring it in the event of a problem. As the OP indicated.
Or...
Push the Exchange server and it's data into a "cloud" provider. Now the clients access the data from the Exchange server in the "cloud" and the "cloud" provider provides DR copies of that data at their network bandwidth to their correctly managed data centers. The cloud provider could manage that Exchange server on your behalf or just provide the infrastructure.
Now when some disaster strikes the DR is performed in the "cloud", at local speeds.
I.e. Why have any local services at all? [assuming security is covered elsewhere... a traditional challenge that exists even in internal datacentres where the local internal admin can access data they shouldn't be able to and backups they shouldn't be able to]
If you're using your applications entirely in the cloud suddenly it all looks a bit of a different problem. How do I get my apps into the cloud, how do I move my apps between cloud providers, how do I ensure my cloud provider is delivering an SLA that is appropriate for my business.
Cloud computing looks like a technology that users really don't want or need.
It's promoted by those eager to turn a one time purchase into a revenue stream. From the seller's point of view, would you rather sell a $100 hard drive, or a $29.95 a month service.
It's supported by the same "hive-mind" of pundits who thought pen-based computing was the next big thing.
Users want control, freedom and low cost.
The "weasels" want a locked-in, never ending, revenue stream.
iDisk is a testament to that...I've been waiting for months to sync a few measly gigs of data using that $99 service.
I think therefore I can't be ~TTNH
"the author spent three months replicating 1TB of home data via cable modem to an online backup service."
What a waste of time and effort. There's a simpler way, but it depends on your provider.
All the author had to do was to set up DRBD on his VM. DRBD supports "truck mode" (as in never underestimate the bandwidth of a truck full of tapes - or USB keys, for you young ones).
Just have the cloud provider set up a USB key, and sync it up with DRBD. Then have the cloud provider Fed-Ex the USB key. Amazon will do this; I don't know about other providers. Once you have the USB key, just sync it back up with DRBD.
I absolutely amazes me of all the bright people who are using cloud services (including PhD's doing research) overlook this simple method.
Save your bandwidth for the updates. Do the heavy lifting with the tools that are out there.
Just buy an 2TB drive and stick in a drawer at work.
I have a friend that signed up for some cloud storage backup and spent months backing up his less than a terabyte. Such a sucker.
We have our backups offsite too. On externally hosted servers that we directly control in a heavily security vetted DC (some of our clients are banks who would demand nothing less even though the backups in question contain non of their operation data aside from emails containing project/spec/contract documents and such) rather than a "cloud" arrangement, but it would still take quite some time to draw the whole lot down over the connection we currently have.
But that isn't a problem because this has been planned for. There are many options to help out here:
Really, if restoring from your backups is a major problem then you didn't plan your backup strategy well. And you probably didn't ever do a test restore before now either (otherwise you'd be prepared for the time/hassle rather than it surprising you) which earns you a "serves your right" slap on the wrists - a backup procedure that does not have a tested restore procedure is insufficient.
Anything above the Economy package has WAY more than enough bandwidth.
Storage networking assumes a symmetric bandwidth pipe. One half of that symmetric pipe uses the bandwidth listed as the maximum possible upload speed -- the number after the /. For cloud based storage to work for a large portion of the connected systems, the Ignorant Lame Egotistical Carriers have to provide significant symmetric bandwidth at an affordable price. I don't see anything symmetric or affordable in what you listed.
Every mans' island needs an ocean; choose your ocean carefully.
I don't see this as an either/or proposition. Backing up protects you from data loss, which comes in many forms:
* Sudden hardware or software failure
* Silent hardware/software failure (or user failure) resulting in corruption you only discover later
* Theft/fire/natural disaster
At the same time you want:
* Easy backup procedure (if it is too hard, you won't do it)
* Fast restore procedure
A sensible backup plan needs to address all of these needs. Incremental tape backups with rotation to an offsite vault is one option which covers most of these things, but isn't particularly easy or automated. RAID is very easy and convenient, but only covers a very narrow range of hardware failures. (If you listen closely, you can hear the screams in the distance of a RAID user who just lost data to software-induced filesystem corruption. Hence the mantra "RAID is not backup.")
Network (blah, blah, "cloud," blah) backup services are a great option for cheap offsite backup that is extremely convenient and continuous. But you should supplement it with some kind of local, fast backup as well. That way you can recover quickly from hard drive failure and corrupted filesystems, but still have a Plan B if your house floods. (Or if you local backup turns out not to be broken when you need it!) Moreover, many network backup services will mail a hard drives for a fee if disaster strikes and you need to restore everything.
In my case, I use CrashPlan and Time Machine to do this. CrashPlan backs up changed files every 15 minutes to several offsite locations. I also plug a Time Machine disk into my laptop periodically to make a local snapshot. Restoration is quick in the common case, but I also have coverage for extraordinary events as well as backups when I travel without my external disk.
How long will your choice of media last even in the pitch dark ? If the data is really worth backing up it is worth restoring from on a regular basis to check for validity. Even long term storage media like mainframe tape only warrants for 12 months and then you need a re-write to 'clean' media. Most businesses satisfy themselves with a 'best-effort' and then just live with the loss. Only those places mandated by strict law or those with a huge potential financial loss ever really deal with the situation.
errr....umm...*whooosh* *whoosh* Is this thing on ?
...was a great hustle we all coulda got it in on w/ eyeOS
That really is a damn unfortunate name for a company, or a subsidiary. I had to read the article carefully to confirm that wasn't a joke.
100% agree. ...and if it was on the "cloud" you wouldn't have access everywhere. Only where the net access wasn't filtered to disallow it.
Plus forget about companies closing down, you'd be at the mercy of the company that now owns your data anyway. If they decided to hike up their rates before you could remove it all, you'd have two choices. Pay up, or lose your data.
Get a 3rd drive though and store a copy of your data off site, updating periodically (maybe once every month or two, or if something you really can't afford to lose comes up).
Cloud is just bad data outsourcing mixed in with thin client. Usually people who go on about how fantastic it is have a cloud to sell ya (and the Golden Gate bridge too, if you're gullible enough).
These posts express my own personal views, not those of my employer
Though I am not an owner of a Sidekick, this is just another in a long line of screwups and bad software by Microsoft.
"It takes months to restore data from a box of harddrives? Sounds like a problem with the backup policy, not the technology."
Considering how we're talking about cloud storage, which would require a good connection able to handle large amounts of data.
At current USA ISP offerings, there is no way in hell to get a fast backup made or restored.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
Oops. I accidentally the whole SAN.
That must be embarassing.
Help stamp out iliturcy.
Many people here sound like the "that horseless carriage is useless" crowd.
Fiber to the endpoint or near the endpoint, with ridiculously high speed wireless for the rest,
will increase. This may be driven by IPTV, who knows, but it is inevitable.
Clouds will become more sophisticated.
They will not be reliant on any single point of failure. Many cloudy infrastructures (like Google)
are already pretty good at that. Much better than your crappy single backup hard-drive.
With luck, clouds will become a layer (stratus?) independent of single hosting companies. Moving clouds.
You can stick with your buggywhips if it will make you feel better.
Where are we going and why are we in a handbasket?
Anyone thinking that the people at "Danger" are rethinking their decision on a company name?
It was an email special. When drive space is that cheap you can have complete redundant backups and store one off site.
I don't see a problem with using a cloud storage provider for redundant off site backup. At least you'd have the data, even if it took a week to restore. If you could prioritize the restore, important and active customers first, everything else later, that might not be all that bad.
That's our life, the big wheel of shit. - The Fat Man, Blue Tango Salvage
After all, you're buying a service from a company called Microsoft/Danger. What could possibly go wrong?
... and then they built the supercollider.
We use both Mozy and Carbonite in our home.
:)
Carbonite has the benefit of backing everything up in real time (a 10 minute delay anyway).
Mozy has the benefit of speed, they backup 10 times as fast as Carbonite.
We use Carbonite to backup our family photos, music collection, documents, etc.
Mozy backs up everything, it took 3 months to backup 4TB of data, but it did the job. Carbonite would take several years to do it because they slow down the more you backup.
The downside to both of course is that if we ever had to restore the data, it would take forever. The upside is that most of that data is not needed quickly. The stuff that would be needed right away could be gotten in a few hours.
It would be nice if either service offered the option to run a backup on DVD-R and mail it to them, then do updates online. However I don't think either service was really meant for 4TB backups for $50/year, even if they do say "unlimited".
---
Cloud Computing Feed @ Feed Distiller
*Gasp!* I wish I could have those speeds at those prices! I have 10/1 for $77.50/mo...
A fool and his lamb are worth two in the bush.
All your data are lost by us.
"We live in a global world" - Harvey Pitt, former Securities and Exchange Commission Chairman
They charge $2.50/GB if you go over your monthly transfer limit. If I lost my data and needed to replace it quickly (assuming I for some reason chose to back up multimedia in the cloud and then suddenly needed all my DVDs at once) it would cost considerably more than buying a highly redundant RAID array.
So if this is the future...where's my jet pack?
You think important business information that people can make money from is going to be safe in the cloud?
You really believe that?
as recent events have shown
to anyone who can out code MS...
From modem using UFO hunters to Russians with adsl, to grandmas with FTTH.
MS failed with your desktop, failed with the net, failed in London, and now you want to trust them with your personal data on the net ???
Domestic spying is now "Benign Information Gathering"
Every time I do the numbers I just can't figure this one out.
It would roughly cost me 20 times to store archive material in the cloud than it would to implement a robust local solution. I haven't bothered to redo the numbers for about 6 months.
There are so many more costs with the cloud.
+ bandwidth
+ monthly space fees
+ registration
And if you don't pay the bill your data usually vanishes.
At the moment SATA drives are extremely cheap. Making backup to physical and then putting the drives in secure storage is very cheap and fast. And I don't loose the data if I forget/can't to pay the bill.
Cloud storage is just a form of data extortion from what I can see. It's like crack, almost free for the first hit and so easy to use. Then you can't get away from it. You want to but it won't let you go. You will fall apart with out if you leave it.
We're reshaping the mobile Internet landscape
Danger is now a part of Microsoft's new Premium Mobile Experiences (PMX) team, a group within the Mobile Communications Business (MCB) of the Entertainment and Devices Division at Microsoft.
But you really cant make the argument that they're preventing HDTV streaming.
Bandwidth cap is 250GB, regardless of upload/download speed or tier purchased. That's 170 hours of footage per month, assuming you don't use it for anything else. You aren't planning on using it for anything else... are you?
#fuckbeta #iamslashdot #dicemustdie
All your pr0n belongs to us.
thirsty.
I dream of a day when we don't need to have expensive computers with lots of processing power and huge storage drives that none of us really understands and which much be replaced every time Microsoft needs to bump up their numbers. Instead all we need is a lightweight 'terminal' with a wireless connection to a larger system that is maintained by professionals who do the backups for us and provide us with all the software and services they have decided we need. The system will be highly reliable and we'll just pay a monthly fee for access plus a small rate per usage unit. I'm not sure what to call such a thing, maybe something like "Mainframe". And we can call our access to it "Time Sharing". And we can then all wait for the day when someone frees us from the whims of those damn time sharing companies by inventing a thing called the personal computer.
Come people... the "cloud"? Been there, done that. Anyone old enough to remember, isn't going to fall for that again.
Since you have so much knowledge about FIRE PROOF SAFES, then how about making sure people know to use a UL Class 125 safe for non-paper media, and not a Class 350, which is for paper?
When Microsoft loses the personal data of its Danger customers, it's not more an indictment of cloud storage than Microsoft Windows blue-screening and losing your data is an indictment of desktop computing.
SpiderOak lets you keep a copy of your backups locally. For $10/mo, I get enough space for my needs. No hassle...as long as they stay in business.
~ Great minds think alike -- so do ours.
I just signed up for Mozy for a measly $54/year. I have almost 9GB of data backed up to their servers that took about a week to completely upload from my laptop when I was occasionally connected to the internet and not using it. I have a very small consulting business and I don't have time to juggle hard drives, run to the bank to keep a secure offsite backup or spend time worrying about my data.
If I don't pay my bill, the data does disappear. So What? I probably moved to a different service or a local backup solution at that point, or my business failed and the backups are the least of my concerns.
You may think it is expensive, but I find it to be a deal. I don't know what it would cost me to replace my data, but it far exceeds the cost in time or money of backing up using Mozy. You may have a different cost/benefit balance sheet and find that these services are too expensive and you may have other reasons you are not comfortable using them. That is fine, but understand your needs are not the same as the millions of people who do find value in online backup.
Having worked in Business Continuity and Resiliency Services for the last 18 months since I graduated college I can assure you that large corporations still prefer backing up to tape, and storing those tapes off site. There is some popularity in keeping off site backups via large pipes to EMC/Shark/FastT/Hitachi/etc. But by far tape is customary and dependable. The problems I have seen with tape libraries are usually quickly resolved. These problems usually boil down to bad tapes (anyone serious would have more than one set), bad tape drives (easily replaceable), robot problems (easily fixed or replaced), SAN infrastructure problems (have people at the off site/hot site facility familiar with cisco fabric manager, SAN switches, fiber, etc.), and software configuration issues (netbackup for example, but for this example anyone serious would have support contracts with Symantec and quickly resolve their configuration issue during recovery).
What is slashdot?
as much as you do. As that is where the *real* money is. Good data is as good as gold, or better.
A hosting company will never understand the value of what you have. And cloud computing is nothing more than that, glorified hosting companies.
Protect your data first and foremost.
putting the 'B' in LGBTQ+
Encryption, amigo. You should look into it.
Absolutely, in what way is cloud storage unsecure?
An SQL query goes to a bar, walks up to a table and asks, "Mind if I join you?"
Send all your DVDs to me. I promise to keep them safe.
An SQL query goes to a bar, walks up to a table and asks, "Mind if I join you?"
This is not surprising. In general, new technical solutions and new business ideas are prone to mistakes and failures because no-one has had time to explore every possible usability problem and failure mode of the new tech+business idea.
Today, some people swear by technology X from company Y. Several other companies have tried and failed, in one way or another, at providing a good product using technology X, but company Y managed to pull it off.
Someday some people will swear by the cloud storage service product A by company Z. Same old story.
The inside of the safe will easily reach temperatures that will destroy your media, but not destroy paper records. Some types of fire safe also contain ablative material on the inside of the safe that is designed to melt onto the papers contained inside to encase and protect them during a fire, which is also extremely bad for digital media. Unless your safe is specifically rated to handle digital storage media in a fire, it most likely will let you down.
Point to the server room, tell them "we've got our own cloud in there and we've already paid for it". Then start talking about the expense of change. They can then go back to their own pointier haired boss and tell them that your company is leading in the cloud and fully buzzword compliant.
It's funny, a few days ago I was pondering how best to speed things up for least expense with solid state disks when a clueless salesperson rang to try to sell me some cloud stuff over a horribly slow link instead which would be like moving back to 9 track tapes for speed. Distributed hosting has it's place but the cloud hype is mostly missing the point. I suppose since nobody cares much about revealing credit card numbers or sensitive medical details of clients then fewer people worry about where their data is kept and who has access to it.
In my experience, if somebody has to do backups, then backups will not be done with any regularity. It's just a fact of life.
Thats why I rsync my approx 12GB of data, stuff that changes all the time, nightly to another machine here in the house, and to a USB drive, then once a week, I do an incremental of the second machine's copy to Amazon S3 using Jungledisk... For what I paid for Jungledisk ($20 one-time) and the recurring costs to Amazon (usually under $2.00/mo, depending on how much more I've uploaded and the transfer/requests charges).. That way, I lose the harddrive on my main machine, the most I've lost is one day, and if the house goes up in smoke, the most I've lost is one week. Jungledisk/Amazon S3 beats the hell out of Mozy/MozyPro/Carbonite, neither of which can run on Linux (Jungledisk *can*).
Spoken like somebody who is truly tech-savvy. So every day you back it up nightly to another machine in the house, and to a USB drive. For 12 GB, even locally, this takes anywhere from 5 to 30 minutes depending on the average file size. So that means that about 200 days /year (if you are PERFECT) you are backing up this data. At 15 minutes per day, that adds up to 50 hours/year of time spent... backing up data.
How much is your time worth?
Let's say you are just a lowly wage slave and earn just $9/hour. 9*50 = $450, which is how much money you're spending per year for your glorious process. I'd guess your time is worth considerably more - probably more like $30/hour. Maybe even $50/hour? In which case, your 50 hours just cost your employer $1500, or even $2,500. That is, if you bother. And I sincerely doubt you actually *are* backing up every day. More likely 1-2x/week, which raises your loss footprint to around 3-5 days/max, which costs quite a bit more.
Yet an online backup service (like Mozy) will back up all that data for $60/year in a way that's almost invisible. It's automatic. It happens EVERY SINGLE DAY that the computer is on. And even non-tech weenies will do it, because all they have to do is INSTALL THE SOFTWARE. Anybody can do this, even a techneophyte. And you can take the other $2,450 and buy one !@#@$! of a computer with it...
This brings the typically techno-centric topic of data backups to the common Joe. And while cloud services aren't perfect, they are usually a damned sight better than the alternatives. Even with something like Mozy (just the 1st name to come to mind, not affiliated, blah blah) the odds of Mozy crashing or dying at the same time that your own computer does is pretty slim, and certainly you are far better off.
I have no problem with your religion until you decide it's reason to deprive others of the truth.
I have 2 USB disks - both Truecrypted.
One of them is at my parents home 20 miles way.
Every day I use robocopy to sync the one I have and every time I visit my folks I swap the disks.
It was a one time investment of 80€ and no subsequent cost.
And I use the free Mozy with 2GB for my frequently changing core data.
Why would I pay anyone a monthly fee to store my data ?
The problems going begging are:
1) Speed of recovery. You have instantaneous access to data backed-up to the cloud. Getting access to your securely-stored hard drive will take longer.
2) Ability to backup-and-forget. Backups to the cloud can be done automatically. You need to physically make and transport manual backups. This is tedious, uninteresting work. People hate doing this kind of thing, so they typically stop trying after a time.
3) Frequency of backups. Backups to the cloud can be done at a much higher frequency than manual backups.
The limitations of cloud-based backups are bandwidth and cost. You would use the cloud only for critical files (such as every file in your home directory tree) where the cost of disruption would be much greater than the cost of backing up.
"We receive as friendly that which agrees with, we resist with dislike that which opposes us" - Faraday
I have a server located in a relatives' house in another jurisdiction and have the computer configured to treat the remote server as though it is locally-attached storage device. Naturally, all traffic is encrypted. The local 250GB external hard drive contains my home subdirectory so in the event of a disaster all I have to do is load the operating system on a new computer and can restore my home subdirectory from the remote server or external hard drive depending upon the type of disaster. I am also looking into deploying eyeOS locally and remotely to provide a true personally-controlled cloud experience with redundancy.
Instead, the cloud has been taken from its' natural setting by companies who want to be for-profit gate-keepers, even though, by their very nature, they will do a worse job (less redundancy, not geographically spread out, etc.)
Oh, you forget all data centres will eventually be located under the oceans for reduced cooling costs and more environmentally-friendly operation. With severals oceans scattered around the planet there will be plenty of geographical dispersion. What could possibly go wrong? Yeah, that silly underwater cable getting severed again thing...
But you should supplement it with some kind of local, fast backup as well. That way you can recover quickly from hard drive failure and corrupted filesystems, but still have a Plan B if your house floods.
Locate a backup server in the floor joists in the basement ceiling and another backup server in the attic. This way your data is protected from loss due to flood (attic is safe) or tornado or hurricane (basement is safe). An off-site backup is still necessary in the event the building is a total loss.
Anybody who accepts and uses the term "cloud" is past redemption. You see there is this property of clouds that applies neatly to the current meaning. Clouds quite often dissipate naturally and quickly leaving nothing but clear blue sky. Not somewhere to keep things safe IMHO.
It seems that every customer I run into with a glitchy backup environment wants to do "online" backup because it requires less investment and the presumption that their data is "safer" offsite. Our occasionally braindead sales often jumps on this bandwagon and I get the virtual equivalent of kicks under the table when I ask about versioning, disaster recovery, data formats, on-site data delivery (ie, all data at once), Active Directory, Exchange, SQL, and metadata recovery. I don't even get into security..
I always get muddy answers when I ask these questions; it may be fine for casual use in the 100GB arena where one only expects or needs the most basic of file recovery, but seems entirely primitive and restrictive when talking larger data sets, disaster recovery, etc.
I'd feel better about these solutions if they would periodically and as requested deliver media in at least the server's native backup system format, or even better, in a local backup system's (ie, BackupExec) native format for sanity testing, versioning, and testing of data integrity.
So you have to physically open the safe every time you want to back up you data?
You can get fire-rated safes with USB pass-through ports:
http://www.google.com/search?q=fire+safe+usb
Haven't used it myself: I just rsync to a FW drive every Sunday evening, and take it to work on Monday (bringing back the drive I have there).
I also have a local Time Machine volume for quick restores.
No, send all your DVDs to me! I promise to keep them safe and I'll even do backups for free!
Can anybody explain to me why it seems to be the general opinion of /. that "the cloud" is a waste of time and a fad? Apart from the silly name, I see many useful functions that can be used for people, who start from a small base, but want to have the ability to easily scale a service (e.g. with Amazons EC2).
If I had a service that is potentially computationally expensive, I could start with a small instance and scale it if I see I need more power or servers. Obviously I could do the same with dedicated servers but I would have more to pay without knowing if I need them and have more difficulties setting them up (I presume). I see it especially useful for small start-ups not so much for large organisations that have the resources to solve these things differently.
What am I missing?
True, I could say, "a model for enabling convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction."
Or, I could say "cloud".
"We receive as friendly that which agrees with, we resist with dislike that which opposes us" - Faraday
"The article makes the argument that high volumes of data and bandwidth limitations make external cloud storage all but useless for enterprises because it could take months to restore the data in a disaster. It also appears to be a consumer problem -- the author spent three months replicating 1TB of home data via cable modem to an online backup service."
I work at a large university, and my group recently concluded writing a 5-year strategy for our storage systems. On the topic of "cloud storage", we concluded that this fits into the overall storage plan, but you need to consider what type of data to put on the cloud.
In our case, we created a matrix of all the different storage available on campus, including our central SAN (4 tiers), HPC storage, NAS, Active Directory storage, cloud, etc. We identified the pros/cons to each storage (and yes, there are pros and cons to each storage type) and gave example usage for each.
We recommended that we leverage "cloud" storage for rarely used data that did not require getting backed up by the university. There is a surprisingly large amount of data that fits this. We recommended against putting any "enterprise" data (enterprise data backup, critical files, etc.) in the cloud.
For example: we have one researcher who generates large data sets from his HPC efforts. After publication, he may need to hang on to this data for 2 years, but may not access it at all during that time. Sure, he can always regenerate the data by re-running his HPC with the same initial conditions (and has done this in the past) but it takes months to complete the run. It takes him weeks to transfer the entire data set to a cloud storage vendor, and weeks to get it back (if he needs it.) When you're comparing weeks to months, for data that doesn't need to be backed up (that's according to the researcher), cloud storage becomes an easy choice.
Another example: we have a vocal faculty who are now claiming that they need up to 1TB of storage, to do with as they like, to help support their informal research needs. Not every faculty claims to need the 1TB, so this is hard to plan for. And those of us in central IT know that most of the researchers who claim "I need 1TB" will only use a few 100 GB at most, making the planning a bit harder still. Assume 1,000 faculty sign up, how much storage do you really need, and how much storage will sit idle? Over-subscription helps address this, but not fully. Again, we looked to cloud storage.
In our report, we recommended the university arrange for cloud storage agreements, just like many universities are doing with GMail, so that we retain ownership over the data we put in "our" part of the cloud. We can establish a service level statement for our consumers of the cloud, so everyone understands the pros/cons and appropriate use - and its limitations.