CLOSE
x

Embed

x

A number of websites became unavailable Tuesday after Amazon’s website hosting service went down unexpectedly. Though the majority of sites affected have since gone back online, some appear to still be facing issues.
USA TODAY

SAN FRANCISCO — It’s really not true that Amazon controls the Internet, though it can sometimes feel like it — especially after Tuesday’s four-hour Internet backend outage that slowed or stopped bits and pieces of tens of thousands of websites from loading.

To paraphrase the German statesman Klemens von Metternich, “when Amazon sneezes, the Internet catches a cold.”

The outage stemmed from a problem with Amazon’s popular cloud service Amazon Web Services that affected a big portion of its S3 system, one of AWS’ storage systems. That ended up being a big deal because Amazon has about 42% of the cloud market by revenue, according to the market research firm Forrester.

That compares to about 15% for Microsoft’s Azure cloud platform and 7% for Google’s, with the rest being taken up by IBM, Oracle and other smaller players.

The S3 system alone is used by 148,213 sites, according to market research firm SimilarTech.

Some groups called out Tuesday’s troubles as an example of the dangers of Amazon’s “monopolistic behavior in our economy,” in the words of the Washington D.C.-based Institute for Local Self-Reliance. Amazon controls too much of the web’s infrastructure, a report from that group said in December.

But the situation is far more fluid than the word monopoly implies, said Dave Bartoletti, a cloud analyst with market research firm Forrester.

While Amazon has done a “fantastic” job of providing these services, other companies are equally good. The cloud marketplace is very much in a rough-and-tumble fight for share.

Microsoft’s Azure service is growing by leaps and bounds and the only reason that Google doesn’t have a larger share of the market is because it was late to the party. But its platform is “rock solid” and customers know it, Bartoletti said.

Perhaps more importantly, the outage made clear to users just how much of the online infrastructure they rely on comes from those people up in Seattle.

Americans know Amazon by the constant flow of brown cardboard boxes it sends to their doors, or the movies it streams to their TVs or the dulcet tones of Alexa as the digital assistant answers their questions on the Echo.

Amazon Web Services is equally important for hundreds of thousands of businesses but invisible to most consumers.

It is, in effect, the back-end to much of the Internet. For sites like Netflix, Spotify, Pinterest and Buzzfeed, as well as tens of thousands of smaller sites, AWS provides cloud-based storage and web services so they don’t have to build their own server farms. This lets them rapidly deploy computing power without having to invest in infrastructure, a huge boon that only very occasionally backfires.

Business store their videos, images or databases on AWS servers and have them served up from those servers to their website. To the visitor, this is a seamless process. To engineers, it’s a complex ballet of files and data constantly shifting across the Internet.

When one of the dancers misses a beat, as AWS’ S3 system did Tuesday, the entire carefully choreographed system begins to falter.

Despite that, the damage itself was actually relatively minimal. Companies didn’t lose data — their ability to quickly access it simply degraded for several hours. Just don’t tell that to the thousands of system engineers who spent Tuesday tearing their hair out.

Elizabeth Weise covers technology from San Francisco. Follow her at @eweise.