Should we run stateful services in Kubernetes or keep cloud ones?

[Madison.A]

I’m trying to decide if we should move our stateful services, like our Redis cache and PostgreSQL databases, into our main Kubernetes clusters instead of managing them as separate cloud vendor services. The idea of having a unified control plane is appealing, but I keep worrying about persistent storage reliability and the operational overhead of managing those volumes and backups ourselves.

[Paul.W]

i tried moving postgres into the cluster and storage complexity blew up fast backups felt flaky and restore times were long we finally kept the database as a managed service for reliability and to keep recovery predictable

[Adam.L]

we ran a small pilot with a redis cache inside the cluster and kept postgres external we tracked uptime latency and backup success rate over two weeks we saved some toil but we spent more time tuning the storage class and the operator than we expected the cost was not obviously lower

[Kyle23]

do you think the real bottleneck is the control plane or is storage reliability the real problem

[JasonZR]

i drifted into this topic while chasing a better backup policy we ended up setting a daily snapshot cadence and a separate archive tier for old data the result was a little calmer ops until a sudden restore test failed due to a misconfigured mount trying to fix it we learned that clear ownership and simple tests matter more than fancy automation sometimes