AWS Outage Halts Everything from Subway Operations to COVID-19 Testing

A widespread AWS outage on Wednesday impacted everything from Prime Music and the New York subway to Ring doorbell systems and COVID-19 testing sites.

AWS attributed the problem to errors with its Kinesis data-streaming service based in Virginia. And while Kinesis does not directly affect most of the services involved in the blackout, it has impacted the infrastructure of 27 other platforms, ZDNet reported.

In fact, the issue got so bad that AWS itself couldn’t keep posting updates to its Service Health Dashboard.

The troubles at AWS clearly illustrated one of the inherent problems with a single-vendor approach to the public cloud. An organization that relies on just one provider is beholden to the capabilities of that brand, even if that company does not suffer downtime on a regular basis.

But the widespread outage the day before Thanksgiving as people prepared for a long holiday weekend and waited in long lines at coronavirus test sites showed how one hiccup can affect everything. New York transit officials couldn’t update alerts on their website because of the AWS outage. One COVID-19 testing provider, Embry Health in Arizona, told people scheduled for nasal swabs that testing was continuing but “much of the work is being done manually.” Employees had to switch from entering data on tablets to registering patients by hand. That resulted in long lines and extended delays.

Multiple Cloud Vendors

The AWS outage represents a key reason why more businesses worldwide are crafting cloud environments with more than one vendor. To that point, Nutanix’s third annual Enterprise Cloud Index indicates that 64% of respondents using public cloud have two or more such environments. They expect that figure to jump to 71% in the next 12 months.

This isn’t an easy task. It takes strategic planning with the help of channel partners. Not all public cloud services have the same capabilities. But a cross-pollination approach to the cloud eases the burden when something like the AWS outage happens.

AWS did not provide an expected time for service restoration.