So AWS was DNS this time?
@madkiwi And just to add, for the non techies.
BGP and DNS are some of the oldest and most reliable protocols on the internet - designed for multiple redundant servers/routers etc. A single server/router failing almost never cause an issue, or if it does it is seconds.
The *problem* is almost always the management systems for these, config changes, and so on, which tend to be centralised and themselves a "single point of cockup", which is a bugger to fix when that breaks *everything*.