Amazon.com Inc.’s main e-commerce site went down for nearly an hour this afternoon for reasons still unknown.
“The gateway page of Amazon.com was offline to some customers for approximately 49 minutes,” an Amazon spokesman says. “Other pages of the site were accessible and (Amazon Web Services) was not impacted.” Amazon Web Services is the unit of Amazon.com Inc. that provides “cloud” or Internet-based computing and data storage services.
The spokesman gave no further details.
The outage began at 1:32 p.m. Central time, according to a spokesman for web monitoring firm Panopta.
A similar firm, AlertBot, confirmed the outage and its general length. “The problem detected was an HTTP503 Error, which correlates to the server not being not being able to respond due to temporary overloading or server maintenance,” says Justin Noll, director of client services for AlertBot. “If Amazon was doing maintenance, I would expect them to put up a temporary page, so I believe the issue was not caused by maintenance.”
According to various reports and comments on Twitter from Amazon users, the problem did not extend to Amazon’s mobile apps, nor product pages that consumers could access from search engines.
“Amazon did recover at (2:25 PM CT),” Noll says. “During the outage a few of our tests did pass with page load times between 10 and 30 seconds, so some visitors may have been able to access Amazon.com. But the website was extremely slow compared to its normal average load time of 1-2 seconds.”
Experts say the problem didn’t seem to stem from a hacking attempt.
“503 HTTP errors generally occur when a server is overloaded or down for maintenance of some sort,” says Jason Abate, CEO and founder of Panopta. “I think it’s likely that some subsystem of theirs that is responsible for rendering the homepage broke either due to a code push (which Amazon does relatively frequently) or some internal hardware failure that prevented that subsystem from rendering the page.”
A code push refers to releasing into production new code from developers. “This introduces new functionality to a site or application, but often times also brings along unexpected bugs,” Abate says.
Noll made similar comments, though he didn’t rule out some sort of action against the e-commerce site. “For Amazon, it doesn’t look like any hacking,” he says. “It could be a network failure/misconfiguration or a large influx in traffic that temporarily overloaded Amazon’s network. Obviously, overloading Amazon wouldn’t be normal traffic so if that is the cause, they may have experienced a denial-of-service attack.”
Amazon is No. 1 in the Internet Retailer Top 500 Guide. According to the most recent edition of the guide, Amazon attracts an estimated 330 million monthly visits, and some 81 million monthly unique visitors. Its site response time stands at 2.20 seconds, down from 2.55 seconds in the 2011 edition. Based on Amazon’s most recent financial report, the e-retailer averages about $7 million in sales per hour.
The Twitter social network also experienced intermittent outages from 7 a.m. through 9:50 a.m. Pacific time on Thursday, the company confirmed on its blog. “This incident has now been resolved. We apologize to users who were affected by this, and we’re working to ensure that similar issues do not occur,” the blog post said.
There was no indication of a connection between the problems at Amazon and Twitter.
Favorite