01 — Context
One operator runs a growing set of streaming, billing, admin and internal services, which have to be deployed consistently across development, staging, on-premise production and cloud production.
02 — Challenge
Many services, several clusters and different hosting locations need predictable, repeatable deployments and shared configuration without drift.
03 — Approach
One declarative repository
A single infrastructure repository of declarative manifests (about 200), organised by environment and then by service group (namespaces, shared defaults, in-house tools, internal services, one group per product).
Build once, promote everywhere
Runtime configuration lives in the infrastructure repository, so application images are built once and promoted between environments.
One way to deploy
A single deploy script targets each environment's cluster. Applications share a common CI template (build image, render manifests, apply).
04 — Outcome
New products are onboarded as a new service group with the same deployment pattern.