ADVERTISEMENT
See Inside Scientific American Volume 310, Issue 2

A Brief History of Epic Cloud Computing Fails

Like the convenience of storing your stuff online? Remember these four crashes, and proceed with caution

As I noted in my Scientific American column this month, storing your data in the cloud is all well and good—if it's optional. Increasingly, though, Apple, Google, Microsoft and others are forcing you to entrust your data to the cloud (their online servers). If you want to use their products, you must agree.

In theory, cloud storage is great. Instead of some $150 hard drive connected to your PC, your data rests on industrial-strength corporate servers, backed up, powered by redundant systems and attached to uninterruptible power supplies. What setup could possibly be safer?

How quickly we forget…

  • MobileMe Becomes a MobileMess: Apple's own MobileMe service (now called iCloud) offered automatic syncing of e-mail, calendar, address books and other data across various Apple gadgets and Macs. But its history was rocky. It erased user data or faced unexpected shutdowns many times during its three-year life: summer 2008 (11 days, resulting in permanent data loss in some cases), December 2009, September 2011 and June 2012, for example.
     
  • Gmail Goes Dark: Millions of people trust their e-mail lives to Google's free, excellent Gmail service. You access it on the Web, so it's available from any computer. Which is all great—until it goes down.

In February 2009 that's what happened. When you tried to log in to check your mail, all you got was a "502 server error." That outage lasted only a few hours, but it wasn't the only one. Gmail had blinked out before, and would blink out again—for five days in 2011, for example.

  • Hotmail Auto-Deletes 17,000 Accounts: It was an errant cleanup script, meant to delete dummy test accounts, that took down 17,000 real people's e-mail accounts at the end of 2010. It took three days for Microsoft to resurrect them from backups.
     
  • Amazon Web Services Melts Down: You might think of Amazon.com as a big online department store. But behind the scenes a huge part of its business is renting computer time, storage space and file-management services to other Web sites—Netflix and Pfizer, for example. This business is called Amazon Web Services (AWS).

But in April 2011 an AWS data center glitch in Virginia triggered a chain reaction. Much of this cloud power went dark on the east coast—for four days. Big-name Web sites including Reddit, Foursquare, Quora and HootSuite went partly or totally dark, and countless smaller businesses could only sit by helplessly as Amazon technicians tried to fix the problem.

PayPal, Intuit (maker of Quicken), Rackspace, Terremark, Salesforce and many other companies have also landed in the headlines when their online services went down.

This isn't an ancient-tech-history thing, either; in 2013 alone Amazon.com's home page went down (one hour), Dropbox was offline (twice—24 hours and 1.5 hours), Facebook stopped updating (three hours), Microsoft's Office 365 online software suite went offline (two hours), the Bing search engine quit finding (two hours), Google Drive vanished (17 hours) and Twitter blinked dark (45 minutes).

The point isn't that cloud services are unreliable; on a percentage basis, they're astonishingly reliable; most run smoothly 99.9 percent of the time, for example. They are not, however, infallible—and they should also be optional.

Scientific American Holiday Sale

Give a Gift &
Get a Gift - Free!

Give a 1 year subscription as low as $14.99

Subscribe Now! >

X

Email this Article

X