Comparison
ControllervsOperator
Controller
you never create the individual containers; you create an object, and a background process creates them to match it.
A process that owns one kind of object and runs a reconciliation loop for it — creating replicas, attaching storage, updating a load balancer. Understanding that there are many of them, running independently and asynchronously, explains why one change ripples outwards over seconds and why the events are the log. Most debugging in an orchestrator is working out which controller is unhappy and reading what it said.
Full entry →Operator
the failover, the backup and the version upgrade for the database are performed by software encoding what the DBA used to do.
A custom controller that manages a stateful application, encoding the operational knowledge for it — how to bootstrap, back up, scale, and upgrade safely — as a reconciliation loop. It exists because generic orchestration handles stateless workloads well and stateful ones badly. The trade is real: you swap a runbook a human follows for software that will act at 3am, and now the quality of that operator is your database's reliability.
Full entry →