Sidekick's lesson: We learn by failing
Technically, we can't blame the loss of Sidekick users' data on a failure of either the concept or the technology of "cloud computing." But Microsoft's clear bungling of basic information management practices (apparently, there were backups--but they didn't work) does cast a pall over not just Microsoft but the cloud concept entirely.
T-Mobile is trying to keep customers' data alive.
(Credit: Screenshot by Rafe Needleman/CNET)Microsoft, as one of the giant infrastructure technology companies that's saying through its product offerings that data is safe in the "cloud," has a responsibility not just to its customers but to the growth of cloud computing overall to keep the data it's holding safe.
The company's failure to keep the data safe shows the world how fragile cloud computing is. Even though, really, it isn't. The world knows how to build systems that safeguard data from hardware and software and network failures, and even from hacking and other forms of sabotage. The fact that Microsoft failed to keep the Sidekick data backed up indicates, rather, how management can fail.
But do consumers, or corporate IT managers considering cloud-based services, care where the failure was? All we know is that it failed.
Travel by commercial airliner is neither unsafe nor inherently safe because of the technology itself. It is as safe or as dangerous as the procedures followed to certify and maintain the equipment that people put their life's trust in.
Microsoft's Sidekick outage shows that sadly, in fact, it's true: you cannot trust the cloud because you can't trust the people who run it. It indicates another scary truth: We haven't had enough cloud failure yet. We're going to have more. We need more. We learn from each failure. And we're all thinking the same thing: I hope I'm the beneficiary, and not the victim, of the hard lessons still to come.
Data in the cloud can be safe. And it will become more safe thanks to this outage. Failures of trust, like this one, have costs, but there are benefits as well.
Rafe Needleman writes about start-ups, new technologies, and Web 2.0 products, as editor of CNET's Webware. E-mail Rafe. 
Frankly, we should stick a fork in cloud computing. We could revisit it in 5-10 years, but it's simply a waste of column-inches right now. These services are being run by a bunch of shoemakers.
Data will be safe if a single drive fails. If the system is writing corrupted data or the system is breached you will not have safe data. If the system replicating backup database info is being send bad data from the primary database, all your data is corrupt across all mirrored drives unless setup correctly.
Obviously this was the case. The database crashed and corrupted the backup when replicating.
RAID 1 (mirrored) will keep identical copies, but performance suffers a bit.
RAID 5 (striped/parity) is the standard on most servers. You can lose one (or two, depending on how many disks you have) and be okay, but you have to let the set rebuild after a failure. Requires at least three drives, which most users do not have stashed in their desktops.
Note that RAID will not save your data if that data gets corrupted (through malware, OS failure, etc), because the corruption will simply be written across the set (and any network drives attached to your device). RAID will not save you from human error - if your kids accidentally re-format the disk, they'll re-format all of the disks in a given set. RAID only prevents a single hard disk failure from destroying your data. That's it.
Only offline backups will reliably save you from most types of failure and mishap. Shadow copies and such are helpful, but if the disk gets reformatted, so do the shadow copies. If the OS gets infected, you can't rely on them to be clean. (hint: get a USB drive, plug it in to your computer once in awhile to run a backup, then UNPLUG THE DAMNED THING when you're done w/ the backup).
RAID only protects from hard-drive failure, nothing more. It does not protect you from corrupted data (bad software, bad controller, bad cables). It does nothing against theft or environmental disasters like flood or fire (you need to store a backup copy offsite).
In 20 years I have lost nothing.
People can keep using "clouds" if they want, but my track record speaks for itself. If you want something done right then do it yourself. It's not computers you have to trust, but the people operating the computers.
Computers don't kill data, people kill data ;)
I might be old fashioned, but I'll stick with how I do things now since I don't have to rely on anyone but myself.
But we have been reminded of MSFT's barking incompetence.
http://www.roughlydrafted.com/2009/10/12/microsofts-sidekickpink-problems-blamed-on-dogfooding-and-sabotage/
Cloud computing architectures can be designed for highly available criteria - and have been, among them Danger, the system that secured the Sidekick customers' data. Until MSFT bought the company and lobotomized it.
Word has it that Ballmer ordered a truck load of chairs to throw at the server farm, hurled everyone of them with that girlie arm toss to no avail.
Duncan Hines,
RSB
Ever heard of a comparable data loss at a large well-known firm? No, you haven't, so this fiasco says nothing about cloud computing, yet confirms what we already knew about Microsoft (managed by morons).
The only reason that you do not hear of OTHER companies having these problems is that they cover it up, say it isn't a loss of data, and their user agreements have severe penalties if you say anything publicly about them losing data.
Other companies have had data losses, but rarely do you see them happen in orgs that store customer-owned data. Microsoft/Danger is the first really big one.
@Lerianis3: Bull Excrement. No NDA can save a corp from being sued over a customer's lost data (even if it's small claims). No NDA can save a corp from having to compensate paying customers for lost data (only an SLA could do that, and in the business world, only a very stupid company would agree to eating lost data w/o compensation).
As for the whole Microsoft Pink thing? Meh - we'll see. It sounds plausible, and Windows Mobile is choking on its own blood right now, but only time will tell.
The truth is, cloud computing is fragile. It takes your data and sends it over an internet connection (driving through 8 mile in a convertible with a suitcase full of hundred dollar bills) and fails intermittently. So far we have seen periodic failures with Google Apps (more than enough to scare most any business or serious user away) and situations such as these where data evaporates.
I doubt we will have this down pat in the next five years. Cloud computing needs time. I'd love to wake up and see technology writers just giving it a break already and go back to talking about virtualization or something else that is truly revolutionizing industry.
The fact is that cloud computing is only fragile IF THE COMPANY IN QUESTION CHEAPS OUT, and doesn't have numerous backup servers to take over when the inevitable failures happen.
Untested backups are a fundamental error, committed by the most incompetent of the incompetent IT.
It has nothing specifically to do with cloud computing.
You can expect more basic, fundamental errors that instead of affecting one company takes out dozens simultaneously as cost competition squeezes competence out.
And its not just the cloud but the communications with it. I don't need my internal business hampered by external problems. We had a landslide last year that took out internet for a week.
We'd be busted today if we were cloud-dependent.
Cloud computing is IT's version of greasy fast food, complete with the e.coli.
That said, there will be those that are gonna wolf it down like the morons they are.
Sadly, it looks like tech journalists are wolfing down the cloud computing fast food like it is going out of style. The last few days have been a god-awful puke-fest. It's almost like these writers are bulimic or something.
Even ignoring the typo, this is pure, unadulterated bunk. Rafe writes as if cloud computing is in its infancy, and there's no way we can hold Microsoft responsible for this unprecedented data loss. Need I point out that RIM has managed to avoid this sort of problem for the last 10 years? Amazon has managed not to lose my purchase history, wish lists, etc. Same with Gmail, Yahoo, and a dozen other high-profile cloud services I could name. Heck, even the two-bit hosting company I utilize hasn't ever lost anyone's data in the 8 years they've been in business.
The fact is, Microsoft screwed up big-time. The attempt to spread the blame over the entire cloud is nothing more than a defensive reflex kicking in, and grasping at the best excuse that came to mind. Pitiful, Rafe.
(That's a rhetorical questions, by the way. I deliberately used the word "unprecedented".)
Several quotes regarding luck come to mind. I'll go with this one: "Luck has a peculiar habit of favoring those who don't depend on it."
But the cloud is operated by people. People make the architectural decisions. They implement it. They maintain and repair it. The cloud doesn't magically create itself.
The SideKick lesson is this: the cloud failed for 1 million SideKick users. Epic data-losing fail. Did anyone learn anything from this? If you are running cloud computing services and don't understand the basic concepts of data backup, you shouldn't be a system administrator. You shouldn't need lessons like this to learn.
Tech writers should stop writing about cloud computing. Give it a rest. It might be worth revisiting in 5-10 years, but then again, it might not. These clowns who are running these services today may be replaced by cheap, even less competent replacements.
First, some other party is holding your data. Trust all you want but back in the dotcom bust, companies were falling out left and right. In a down economy, if the company going out of business runs your cloud, you're screwed. I haven't seen it yet but it's a definite possibility.
Secondly, privacy is an issue. Privacy policies change all the time, usually for the benefit of the company, seldom for the benefit of the consumer.
Finally, control is an issue. This fiasco says it all as far as control.
I understand that this was associated with a cell phone and probably beyond the users' power to decide "to use the cloud/not to use the cloud" but it illustrates the dangers of cloud nonetheless. Not everything that tech companies dream up is a good idea. Cloud computing is the current version of thin client computing, which is old and better suited to companies than consumers. I'm sure I'll be criticized for this opinion in some way but I definitely won't be crying when a cloud blows away and I still have my data.
It seems that cloud computing at least as run by MS is following the same trajectory.
or use a reliable online backup service like Carbonite.
I am continually amazed at how much information people upload to Facebook and backup to some CLOUD with total faith it will always be there. The idea of having your data accessible from anywhere, anytime has it's price.
Image your hard drives at home to an external drive and then go put that external HD in your bank's safe deposit box.
There is a reason the computing paradigm moved away from the mainframe/client model decades ago. None of my data resides solely in the "cloud". Who came up with that name anyway, it's a stupid name. I will always entrust my data to only one person, me.
Sidebar: although the treo did, it always was a painful experience.
If you can backup from a sidekick locally, please advise. And if it's truly easy, then I agree with this comment. Customers should keep their own backups. Although the demographic for a sidekick user is pretty much a teen, I doubt they'd backup even it came with a cable and an 'easy' button to push for backups.
Jerry: I don't understand. Do you have my data?
IT: We have your backup, we just can't restore it.
Jerry: But the backup keeps the data here, that's why you have the backup!
IT: I think I know why we have backups.
Jerry: I don't think you do. You see, you know how to MAKE the backup, you just don't know how to RESTORE the backup. And that's really the most important part of the backup: the restoring. Anybody can just make them.
- by Mikebanks October 17, 2009 3:36 PM PDT
- Damn--when I saw all those "Sidekick" headlines I had a flashback to DOS (Borland Sidekick).
- Like this Reply to this comment
-
(46 Comments)--Mike