Trending: NYC Council grills Amazon over HQ2: ‘You’re worth $1 trillion. Why do you need our $3 billion?’
The Amazon Day 1 building is named after the Jeff Bezos philosophy that every day should still be “Day 1” at the tech giant. (GeekWire Photo / Kaitlyn Wang)

Even though it has the world’s leading cloud-computing service within the corporate family, Amazon engineers have to configure its retail web site just like everybody else. After a series of glitches marred the first few hours of its annual Prime Day promotion earlier this week, it looks like they failed to anticipate just how much incoming traffic would arrive.

CNBC reported Thursday that it obtained internal documents of Amazon engineers discussing what to do shortly after the official launch of Prime Day at noon PT Monday, as traffic began to overwhelm the site and cause issues with product searches and shopping carts. According to the report, the main problem was a breakdown in a piece of internally developed technology called Sable, which retailers and partners selling goods on Amazon use for data storage.

Amazon responded by stripping down the graphics on its home page, cutting off international traffic in hopes of slowing down the fire hose, and racing to add server capacity as the problems with Sable spread to other parts of its architecture. Theoretically, it should have had auto-scaling capabilities lined up, but it also appears something went wrong there; individual people shouldn’t be (literally) running around trying to find servers at one of the biggest technology companies on the planet in 2018 on one of its most important days of the year.

The site began to recover as Amazon added servers, and Prime Day went on to be “the biggest shopping event in Amazon history,” according to an Amazon press release issued Wednesday. The company didn’t say anything about revenue related to Prime Day, part of its usual strategy of serving up very selective numbers, but it had to have left some money on the table during the first few hours.

The CNBC report says nothing about Amazon Web Services, which implies, in line with what an informed source told GeekWire on Tuesday, that configuration errors on Amazon’s end were to blame for the lack of server capacity. Representatives from Amazon declined to comment to CNBC, and did not immediately respond to a request for comment on the report.

Like what you're reading? Subscribe to GeekWire's free newsletters to catch every headline

Comments

Job Listings on GeekWork

Find more jobs on GeekWork. Employers, post a job here.