Interview with Compose

Compose, a database-as-a-service provider, transitioned from Graphite and InfluxDB to Prometheus due to issues with scalability, data corruption, and unfulfilled feature promises. They found Prometheus surprisingly efficient and scalable – currently monitoring over 400,000 metrics across 450 hosts and 40,000 containers with a single instance – and have seen improved reliability and team enthusiasm compared to their previous solutions. Compose plans to expand Prometheus monitoring to include database-level stats and integrate more infrastructure components, anticipating future sharding as they continue to grow.