The question about good practice for fdb cluster

we have deploied FDB clusters by fdb-kubernetes-operator in test environment, default processes include 1 controller, 4 logs and 3 storages.
The test went well, so we plan to expand the cluster to 32 nodes to test with TB datas, what is the recommand of cluster components.
I mean how many coordinators, storages, logs or other processes we need to deploy.