How a thunderstorm knocked out Instagram, Pinterest, and Netflix

Plenty of people found themselves bored this weekend when a powerful Atlantic storm disrupted Amazon's cloud storage centers Friday night, taking down Instagram, Pinterest, and Netflix, among others. The outages, which sent Amazon into damage control mode for the better part of the weekend, underscore how dependent a large chunk of the web is on the online giant to provide affordable cloud storage. Here, a brief guide to the Amazon architecture that provides much of the web's backbone:

What kind of service does Amazon provide?

The Seattle-based e-tailer runs a data center called the Elastic Compute Cloud (EC2) in northern Virginia, which is home to the physical servers that many popular websites and apps use to store their information. When you visit one of these websites, you're actually accessing the information physically stored at this center in Virginia.

Subscribe to The Week

Escape your echo chamber. Get the facts behind the news, plus analysis from multiple perspectives.

SUBSCRIBE & SAVE

Sign up for The Week's Free Newsletters

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

Sign up

And what happened?

On Friday night around 11 p.m. EST, severe thunderstorms cut off EC2's power, crippling services as diverse as Netflix, Pinterest, 99Designs, Active.com, Heroku, and Instagram. Other properties, such as Foursquare, were only partially affected, and most of the companies took to Twitter to alert their customers that they were working to restore service. By Saturday morning, most were back online again, even if with limited functionality. This weekend's outage, however, marks the second time EC2 has experienced downtime in a month.

How bad were the storms?

About 2.1 million people from Virginia to Illinois were still without power as of Monday morning, says Scott DiSavino at Reuters. The storms claimed at least 15 lives, "mostly from falling trees and branches." Ohio-based American Electric Power (AEP), which supplies power to 5 million customers in 11 states, says it could take a few days to restore power to all, leaving millions without air conditioning as the region experiences a heat wave.

Why do all these companies rely on Amazon?

Amazon's EC2 storage is affordable, which is why so many companies use it so extensively. In fact, "many new companies have business models that only make sense (if, in fact, they do) because of how inexpensive and flexible Amazon's cloud services are," says Anthony Wing Kosner at Forbes. But Amazon's vulnerability in such a disaster-prone area could spell bad news, especially if its clients begin looking into competitors, says Sean Ludwig at VentureBeat, "including Rackspace, SoftLayer, Microsoft's Azure, and Google's just-introduced Compute Engine."

Could this happen again?

Unfortunately, yes. The web is vulnerable to many kinds of issues: Not only did Friday's storms cause huge problems, but on Saturday night, after an extra second was added to Earth's universal clocks, a "leap second bug" affected sites like Reddit, Yelp, and LinkedIn. Outages like this weekend's are "a reminder of how fragile, still, our digital architectures actually are," says Megan Garber at The Atlantic. "As one commentator put it, 'No matter how powerful we become as a species with our technology, we are still at the mercy of the clouds.'"

Sources: The Atlantic, CNET, Forbes, Reuters, VentureBeat