When Amazon Web Services Goes Down, So Does a Lot of the Web

Illustration: Konstantin Sergeyev

If your Alexa suddenly went silent or your Slack channel at work stopped working today, there was a reason for that: One of Amazon Web Services data centers in Virginia suffered downtime from roughly 9:23 a.m. to 1:26 p.m. ET today, affecting customers up and down the Eastern Seaboard. And if you think you aren’t a customer of Amazon Web Services, you’re probably wrong — you just don’t know it.

Amazon Web Services (AWS) is a suite of cloud-computing services that, unless you run a tech start-up or pay close attention to Amazon’s earnings reports, you may never have heard of. But the Seattle-based company has quietly become responsible for keeping much of the internet running.

The quick story of AWS: In the early 2000s, as Amazon shifted from “a company that sold you books” to “the company that will sell you everything,” Amazon decided that instead of having each product team build its own servers and databases, it would centralize everything for the entire company. Engineers could remotely access servers that would allow them to get at the computing, storage, and database needs any individual project would require. In 2003, Amazon realized that it had, without quite meaning to, become very good at managing the then still-nascent idea of running remote databases and servers, what would come to be known as “cloud computing.” It also realized that it could sell that service.

Quietly launched as a side business in 2006, AWS was a simple proposition that hit at exactly the right time. It allows anyone, from random individuals to tech start-ups to billion-dollar companies like Slack, to offload the need to run and maintain servers. It controls a huge chunk of the cloud-server market — about 40 percent in mid-2017, per Synergy Research. (By itself, it controls more of the cloud-computer market than its three closest rivals, IBM, Google, and Microsoft, combined.)

If you use Netflix, Pinterest, Airbnb, Slack, or any of Adobe’s web services, you’re indirectly using AWS. And, of course, you use AWS anytime you use any Amazon product, whether that’s Alexa or Amazon Video. New York (and many, many other media publishers) use AWS. In general, cloud computing, as pioneered by AWS, has allowed for the tremendous shift in how the internet behaves and feels — why everything feels like a piece of software, even if very little of it is actually stored on the physical device you’re using.

But that also means that when AWS suffers downtimes, suddenly a big chunk of the web you either rely on for work or rely on for distraction are also affected. In 2017, its S3 (Simple Storage Service) servers went down in February and again in September, resulting in many, many websites just outright breaking.

In this case, it was only AWS Direct Connect, which basically serves a massive, dedicated connection between AWS and client companies, which mainly hit larger enterprise clients like Slack, the bank Capital One, Atlassian Bitbucket (though it also meant that Alexa suddenly was unable to respond to queries from people on the East Coast). In your daily life, that may have meant that Slack was buggy for a bit, or Alexa stopped responding. But it should also act as a good reminder: Much like the electric grid or cell phones, AWS has come to underpin so much of our daily life that we hardly even notice how important its become — until it stops working.


When Amazon Web Services Goes Down, So Does a Lot of the Web