While Apple faces the mother of all iPad backlogs, Amazon.com may be facing the mother of all cloud outages. Indeed, it's disruptive when Google's web services go down. But when Amazon Web Services goes down, it sends a ripple across the Internet.
Amazon was still working Friday to restore the computers that other major Internet brands rely on to serve customers. The outage is bringing attention to the fact that Amazon does more than sell books -- it powers sites like Foursquare, Twitter, Netflix and Zynga. These and other companies rent servers that run their online services.
Amazon's woes -- and the subsequent woes of those who depend on its Elastic Compute Cloud (EC2) -- are giving critics of cloud computing more ammunition against the movement. At the least, Amazon's rare but severe outage could hamper its reputation for dependability.
The Ripple Effect
The cloud drama began in the wee hours of Thursday morning. Amazon's Service Health Dashboard reported connectivity issues impacting its Relations Database Service. The latter is used to manage a relational database in the cloud that spans multiple time zones. Those problems spilled over into Amazon's EC2 service, leaving social-networking sites like Reddit in the dark. (At 2 p.m. Eastern time Friday, Reddit's site was in "emergency mode.")
At 9:18 a.m. Thursday, Amazon offered an update: "We're starting to see more meaningful progress in restoring volumes (many have been restored in the last few hours) and expect this progress to continue over the next few hours." Amazon offered another update at 11:49 a.m.: "Our current estimate is that the majority of volumes will be recovered over the next five to six hours."
Calling for Backup
As Rob Enderle sees it, the Amazon EC2 outage is simply a dramatic reminder that public cloud services aren't particularly robust. Indeed, public cloud services are still risky. The lesson, then, is this: Any function or service that's approaching mission critical needs to have a fall-back plan in case the service goes down.
"Amazon is considered one of the most robust of the public cloud providers; the fact that they went down and then couldn't come up right away has cast a very big cloud over this," said Enderle, principal analyst at the Enderle Group. "A lot of us are now calling for companies that are using services like this to have redundancy programs in place so that if the service fails, you can fail over. Much like you would do if you had an unreliable electrical vendor. You'd put in generators."
While there is no rush to abandon the cloud and its benefits, the belief that the cloud is close to bulletproof has been shattered. Gmail or Google Apps outages may be inconvenient, but the failure of Amazon EC2 is akin to the failure of a hosting service.
"In theory, these have so much redundancy that they never fail," Enderle said. "The fact that this one did really put the question of how robust anybody has made these public clouds."
Posted: 2011-04-23 @ 8:11pm PT
The cloud hype is so overblown. Virtualization and co-los are almost always a more secure, private, cost effective and reliable solution.