We had degraded database performance for a 4 minute period from 08:50UTC to 08:54UTC. During this time, some requests were timing out. The service recovered automatically and has been fully operational since.