Cloud computing poster child Amazon.com suffered a significant breakdown on Tuesday that lasted for hours and affected core e-commerce functionality on its main site.
As acknowledged on its official Amazon Seller Community discussion forum, buyers couldn't place orders while inventory management features malfunctioned for merchants.
Amazon declared the problem solved about eight hours after first acknowledging it, making the disruption a major one for one of the world's largest and most popular e-commerce sites, for which even minor outages translate into significant lost revenue opportunities.
The worst period occurred between 2 p.m. and 8 p.m. U.S. Eastern Time, when 42 percent of all visitors were unable to complete a simple process of getting to Amazon.com, finding a product and placing it in their shopping cart, said Gary Beerman, vice president of marketing for AlertSite, a Web performance and uptime monitoring company.
"That's pretty dramatic," he said.
Problems included pages that didn't load at all or loaded only partially. During this time, Amazon.com served up generic error messages, instead of acknowledging it was dealing with an extraordinary problem, Beerman noted.
Instead, Amazon.com should have posted prominent notifications about the technical issues, ideally with an estimate for when they would be fixed, so that shoppers would have been aware of the situation, he said.
"When you have a problem, it's better to be pro-active and acknowledge you're having an issue than try to sweep it under the rug," he said. "Otherwise, people are just going to keep trying and getting increasingly more frustrated."
When frustration peaks, shoppers give up and go to a competing site to buy the desired products, Beerman said. However, if they know there's a special problem on Amazon.com and that it's being worked on, they might pause their shopping and come back later, he said.
The incident is also significant because in recent years Amazon.com has become one of the most enthusiastic proponents of cloud computing, urging companies to rely on its systems to handle their IT infrastructure functions and tasks.
Through its Amazon Web Services division, the company provides a broad variety of hosted computing services for companies that decide it's cheaper, more convenient and more reliable to entrust them to Amazon than to do them in-house.
Those hosted IT infrastructure services were apparently unaffected by the problems that hit Amazon.com's site on Tuesday.
Amazon didn't immediately respond to a request seeking comment on what went wrong and what impact it had on its sales and those of merchants that use its platform.
This type of problem is more common among e-stores during big gift-giving days and seasons, like Mother's Day and Christmas, but Tuesday was just a regular day in that respect.
"This must have been caused by some technical problem, not by excess demand," Beerman said.