Beyond Built-in Engines: Extending OpenEverest into a Universal Database Control Plane

Running a DBaaS platform on Kubernetes means dealing with a fragmented operator ecosystem — each engine ships its own Operator, CRDs, backup and monitoring setup. No shared API. Platform teams write separate integration code per engine, maintain separate runbooks, and re-solve the same problems each time a new engine is added.

This session walks through how the presenters built a unified control plane on top of OpenEverest — a custom adapter layer with CRD translation, a lifecycle bridge for engine-specific behaviors, and a status aggregation layer normalizing everything back to a single DatabaseCluster API. Any engine with a Kubernetes Operator can be plugged in.

In the staging environment, multiple engines across different workload types are provisioned through the same pipeline — same backup, monitoring, upgrades, no engine-specific tooling. The talk covers what the adapter looks like in practice, where it got complicated, and how Day 2 ops work across a heterogeneous fleet.