On September 4 at 13:41 UTC, the end-user latency across Mews products increased and stayed elevated until 13:59 UTC.
We deployed a change that led, by updating a high-traffic table schema, to an excessive database utilization.
We rolled the deployment back, which resolved the issue.
We identify opportunities to avoid doing the problematic database changes in some situations and to mitigate their impact in situations where they need to be executed.