SAN FRANCISCO, CALIFORNIA
Numerous web sites and apps went down for several hours February 28 due to a typo. Technicians at Amazon Web Services were debugging an issue bogging down transaction times for its Simple Storage Service, which offers free web storage to developers. Amazon accepts technical issues as a part of life and has thus built its system so portions may be taken out of commission without disrupting service. But this time, somebody accidentally took down a seriously large number of servers. Among affected sites were the US Securities and Exchange Commission’s, an Amazon subsystem that indexes all metadata and its storage location for the region, and a second subsystem that places new data and relies on the first. (Also affected: digital publishing platform Issuu, which hosts the digital version of Capital at Play.) Because even the dashboard that tells developers which Amazon services are operational was affected, updates on the outage were posted Twitter. Amazon says the problem won’t recur. It has implemented stronger error trapping, is expediting plans to downsize data cells, and is auditing other processes to prevent analogous errors.
