Amazon S3 outage (2017)
A four-hour interruption of Amazon's storage service in its main region, caused by a mistyped maintenance command.
Object storage is the layer on which a large share of cloud-hosted sites and applications rest, and a single cloud region concentrates a considerable part of it. During a maintenance operation, an engineer entered a command that removed far more servers than intended, including those of two subsystems the whole service in the region depended on. Restarting them fully had not been done in years and took several hours. The dashboard that displayed service health itself depended on the failed service, and it kept showing normal status for a long time. For insurance, the event is the textbook case of human operational error at a provider, with no malice, hitting thousands of dependent insureds at once: exactly the non-malicious accumulation scenario that cyber models struggle to price.
At 9:37 a.m. Pacific time on February 28, 2017, a mistyped command took S3 down in the us-east-1 region for about four hours; the modeling firm Cyence estimated losses for S&P 500 companies at 150 million dollars.
Follow the research
One letter a week. First, the week's underwriting note, when one is published: a dated, argued decision, signed by Paul Bernardski. Then what the observatory published on cyber risk, political risk and insurance. No account, no tracking.
No spam, one-click unsubscribe, no data ever sold.