Amazon’s AWS cloud fails and the IT world (not) stands still

From 9:37 a.m. to 4:35 p.m. PST, at least parts of the US-EAST-1 AWS region of the Amazon cloud failed on Tuesday. Depending on the location of the observer, this almost seven-hour disruption had drastically different consequences: While nothing seemed to work in the USA at times, the rest of the IT world was far less affected – at least when looking at the major services.

At the top of the list of missing: Netflix, Disney + and Amazon Prime. According to several reports, video streaming no longer worked in the United States, but it still worked in Europe. In the USA, however, the effects on Amazon’s own delivery service seem to have been much more serious: This relies on internal apps such as Flex and A to Z, without which warehouse workers, messengers and other employees cannot work – and the parcels piled up. Other Amazon offers such as the Ring security system no longer worked.

Even after the failure has been officially rectified, it is not clear how many services from which other providers were affected, where and to what extent. Furthermore, it is questionable whether the scope of the disruption can be easily surveyed: Amazon itself only made such reports after initial reports on forums Reddit and Hacker News published in the AWS Service Health Dashboard that several cloud services on US-EAST-1 would not be accessible through APIs. After three hours it was clear: The reason for this was “impaired network devices”.

In the meantime, the failure has officially been completely resolved. But technical details about the background are still missing.

(Bild: Screenshot AWS Service Health Dashboard)

The US-EAST-1 region is so important that serious economists even fear a recession in the event of a medium-term failure. EC2, Connect, DynamoDB, Glue, Athena, Timestream and Chime were definitely affected – among other things, as Amazon itself states. And above all, customers around the world could not log into the AWS console, because the global version is also at home here. Only switching to one of the regional variants helped.

Apart from the reports, it will remain unclear how many internal company applications have failed. According to reports from several market researchers, Amazon is by far the largest cloud provider, especially in the infrastructure sector. Correspondingly, IT in many companies behind the scenes depends on this one provider – but, as has now been shown, contrary to the importance and fears, at least not globally on one region.

Regardless of this, the disruption is part of that of Cloudflare 2020 and Facebook in the middle of the year: the increasing interlinking of IT and the power of individual providers are increasingly causing unexpectedly large failures. It is unclear whether Amazon will provide an article on the technical background that goes beyond the short and vague report on the AWS Service Health Dashboard – at least in the past, however, the provider has followed up on its blog.

More from iX magazine

More from iX magazine

More from iX magazine

More from iX magazine


To home page