Tuesday’s internet outages were quickly discovered to be caused by outages of Amazon’s Simple Storage Service (S3), cloud servers built to hold enormous portions of the web.
The root of those outages was uncovered by Amazon yesterday: An employee made a typo while trying to remove a “small number” of servers from S3, accidentally removing far more than intended. The system could not could not prevent the typo from removing two crucial servers which acted as a catalogue for finding all existing files, and a mechanism for adding new files.
Amazon then had to restart its systems— which it had not done for years— a process that took more time than originally expected.
Outages were estimated to cost S&P 500 companies $150 million, and $160 million for financial services using S3, according to Cyence, although there’s no indication Amazon is on the hook for these losses.
Amazon has already modified its S3 system to avoid this problem again in the future.