On March 11th, one of our internal services became unavailable between 12:43 PM to 13:00 PM (PDT). It resulted in increased errors and/or timeouts on those API calls that depend on that service.
A recent change caused inconsistent higher latency of our internal services that handles map geometry. This resulted in some of our customer API requests not being served or experiencing a longer than usual wait for a response.
We shipped a fix to lower the latency to the internal service. Also, we have started the work to set up more monitoring and alerts to our internal services so that we will know the issue before it causes a negative impact.