The database that backs our queuing service encountered a spike in operations and with response latency, leading to a backlog of requests and heightened latency with queue-related operations (including labeling). We were able to scale our services and reduce latency from these requests, which cleared out the backlog sooner, and operations resumed. We’ve made some configuration changes to the database, more granular monitoring and alerting, and have updated the database size and scale itself to handle a greater load.