E-commerce giant's server outage hits web hard

A key Amazon service that helps power the internet went down yesterday, bringing with it scores of websites and services, including Amazon’s own network status page.

“A lot of people have put their stuff on Amazon, so that means when the infrastructure breaks, which doesn’t happen very often, lots of things break,” said Jim Waldo, a professor and chief technology officer for Harvard’s John A. Paulson School of Engineering and Applied Sciences.

Amazon Web Services S3, a server offering from the online shopping giant, had “high error rates” in the East Coast region, harming at least parts of many websites, according to Amazon. Some sites rely on S3 for critical components, while others only use it to access images.

Amazon’s own webpage, designed to show whether AWS services are working, also uses S3 to store its status images, so the page showed healthy, green checkmarks for hours despite the outage.

Amazon said it resolved the issue at about 5 p.m. yesterday, and said earlier in the day it believes it has identified the cause.

The outage affected websites including businessinsider.com, Giphy, Quora and even downforeveryoneorjustme.com, a website made to check if a page is experiencing an outage.

Waldo said the outage is unusual because cloud servers have become increasingly reliable.

“In some sense, Amazon is a victim of their own success,” Waldo said. “They’ve become so successful that people have become so used to their computing services being there that it’s notable when it’s not.”