A host in the region became overloaded, and caused degraded performance on sites running on that node.
Sites running on the affected node would have seen performance issues or timeouts from approximately 14:20 UTC to 16:10 UTC. Some sites experienced a number of brief outages during those times.
An additional host was added to the region, and some projects on the overloaded host were migrated to the new host. This balanced the load issues and resolved the performance problems we saw.