|
|
|
|
|
|
by __turbobrew__
117 days ago
|
|
|
The way I think of it is that a DB should only be accessed by a single replicaset in k8s. Only processes of identical code should share the DB. Everything else is through RPC interfaces. This is how large scale systems are built, but the pattern makes less sense the smaller your footprint is. |
|
|