On Friday night, Amazon’s Elastic Compute Cloud crashed as a result of inclement weather in North Virginia, where the cloud servers are housed. The storm that took down Amazon’s cloud had winds which reached 80 mph and left an estimated 400,000 people without power in the Washington, D.C. area. Popular websites Netflix, Instagram, and Pinterest, all of which run on this EC2 cloud, were among those affected by the crash, in addition to Heroku, a platform site which runs a large number of websites.
The sites crashed at roughly 11:10 PM EST, and Netflix remained down until around 1:15 AM EST, while Pinterest remained down until 1:50 AM. Both Netflix and Instagram posted to their customer service twitter accounts during the outages that they were working to restore service, while Amazon did the same on its service dashboard.
Two hours without Netflix may seem like a nuisance that is hardly unbearable, but this crash of Amazon’s EC2 cloud comes barely two weeks after an outage that lasted six and a half hours. Combined, these two crashes raise serious questions about the ability of cloud computing to withstand severe weather and the risks that come from hosting many popular websites in one location. Although Amazon claims to have redundancies in place to keep the cloud active even when one location faces a power outage, these two crashes suggest that the technology may not yet be at the necessary level to shoulder this responsibility.
As companies scramble to lead the future in cloud computing, the problems with such extensive consolidation of servers will continue to become apparent. When one company is placed in charge of hosting a number of websites, they all become more vulnerable to power outages caused by unavoidable weather conditions. I shudder to imagine a world in which consolidation continues until every website we use is hosted on one of only a few clouds, and a simple thunderstorm has the power to take down the entire Internet. That may be a hyperbolic prediction for the future, but the point still stands that as the role of cClouds to the Internet increases, so too will their technology and security measures be forced to increase.
Forbes called Friday night’s crash “a small step backwards, perhaps, for cloud computing,” but I believe the opposite to be the case. By displaying the flaws still inherent in Amazon’s cloud technology, this crash has both provided Amazon with a chance to fix their week points and has reminded tech companies and consumers alike that the race to be the biggest cloud is not yet won. There is no better motivation for progress than competition, so a reminder that the market remains open to a reliable Cloud should lead to better technology.