Description
Description
Once we fully switch to MeshService, it seems unnecessary to generate Dataplane resources on Kubernetes. We primarily use Dataplane for Universal deployments, where we don't have a way to pass configuration to kuma-dp. I believe we should continue using it for Universal, but remove it on Kubernetes and configure the sidecar based on annotations/labels and MeshService. This should reduce the number of requests to the database. Also, we won't need to sync Dataplane resources anymore and can rely solely on MeshService.
It probably requires a more thorough review, but I feel like we don't need to store this information — we could just build it on the fly instead.
We can build it during connection time based on the pod and store in memory shared over all components. Based on this we could generate dpp configuration instead of pooling db.
Feel free to add more information since I might be missing something.