Amazon Cloud Outage Hits Customers Including Roku, Adobe
(Bloomberg) -- Amazon.com Inc.’s cloud-computing division suffered an outage on Wednesday that affected several customers, including Roku Inc. and Adobe Inc.
Amazon Web Services’s status page noted that its Kinesis data streaming service was “currently impaired” in the company’s U.S. East 1 region. The outages were also making it harder to post updates to a closely watched status page, the company said.
“Kinesis has been experiencing increased error rates this morning in our US-East-1 Region that’s impacted some other AWS services,” a company spokeswoman said in an emailed statement. “We are working toward resolution.”
AWS is the largest provider of rented computing power and software services, and its data centers serve as the invisible foundation of much of the internet. That gives failures in its services an immediate visibility that rivals like Microsoft Corp. and Alphabet Inc.’s Google sometimes don’t face.
AWS is a collection of more than 175 software services, from data storage to a range of databases and machine-learning software. Customers often use more than one, linking them together in ways that can cause a failure in one system to cascade across multiple programs.
The Seattle-based company operates those services from 24 regions, or clusters of data centers, geographic redundancy designed to station computing power close to customers while limiting the chance that a failure in any single region will result in permanent loss of data. U.S. East-1, which relies on data centers clustered in northern Virginia, is among AWS’s most important regions, analysts say.
Kinesis Data Streams, the service at the root of Wednesday’s outage, captures and performs analytics on data, including social media feeds, dumps of public records and internal application usage logs, which can be then be fed into a variety of other software programs.
Jaspreet Singh, chief executive officer of Druva Inc., a data backup and disaster recovery software maker that uses AWS services, said his engineers first noticed the outage early Wednesday morning when the flow of notifications from an AWS data monitoring service were disrupted.
While the outage didn’t completely sever access to a critical AWS service, it seemed to touch more products than previous outages, Singh said.
“Typically what tends to happen is one service goes down” for a half hour or so, he said. “This is a different kind of issue. It’s bigger. Things are failing internally.”
©2020 Bloomberg L.P.