On September 14th, between 17:22 and 18:18 UTC, traffic to our us-west-2 region experienced slower response times and eventually started returning timeout responses.
This was caused by instability with the underlying infrastructure for our database replica in the region causing significant slowdowns in queries being processed, leading to longer times to service requests, which caused multiple web server instances in the region to be replaced, amplifying the problem.
The situation was resolved by working with our infrastructure provider to route all us-west-2 traffic to our eu-west-1 region, route the problematic database replica in us-west-2, and then route the affected traffic back to us-west-2 after recovery.