Suppose I have a local desktop PC with 32 CPU and 128GB memory, and I want to install SingleStore 7.x locally for testing and development purpose with a Free License. I can think of two configurations:
-
option 1: use the cluster-in-a-box docker image, which has 1 master-aggregator node and 1 leaf node. I believe the leaf node uses 8 partitions by default. I can increase it to 32 to match the CPU count so that all CPUs can be utilized
-
option 2, use k8s or minikube to deploy a 4 leaf node cluster, each leaf uses 8 CPU and 32GB memory. Add another node as master aggregator.
Which option has better performance?