Fastly outage of June 8, 2021
An hour of outages on major websites on June 8, 2021, triggered by one customer's valid configuration at the Fastly delivery network: accumulation through a common provider, in miniature.
On June 8, 2021, at 09:47 UTC, 85% of Fastly's network returned errors. Gov.uk, the New York Times, the Guardian, Reddit, Amazon and hundreds of other sites went offline at the same time. Fastly detected the outage within a minute and restored 95% of the network within 49 minutes. The cause, published the same day, was almost trivial. A customer changed its configuration, in a perfectly valid way; the change triggered a bug introduced in the network software on May 12. One action, one customer, one hour of outage for part of the web. The insured loss was small, because the duration was short and most policies carry a waiting period in hours, often eight to twelve, before paying business interruption. The case matters for what it shows about accumulation. A content delivery network is a common chokepoint the insured does not see, since it often has no direct contract with it. A cyber modeler looking for provider outage scenarios starts from that hour, and multiplies it.
Hypothetical: an online news publisher loses an hour of audience and advertising on June 8, 2021, worth 40,000 euros. Its cyber policy covers the failure of an unnamed provider, but after an eight-hour waiting period. The claim is not even filed, and the broker uses it to ask for a four-hour waiting period at renewal.
Fastly outage, Fastly 2021, Fastly CDN outage
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.