Hi I am looking for some guidance on resource allocation for setting up a causal cluster with read replicas.
We have been experimenting for some time with a single server instance, and have reached a point where we need to scale, due to data size.
Data size is:
330GB On DB data
225,000,000 nodes
1,180,000,000 relationships
(yes this is a large database)
At the moment we have it running on single server, with following configuration:
24 cores
256Gb memory (31G heap, 195Gb page-cache)
After looking at the documentation for setting up a causal cluster, I'm still unsure on what memory capacity would be required and how it should be split across each server in the cluster?
My thinking is a setup with:
3 Primary Core servers
2 Read-Replica servers
However, what memory should we have for the primary core servers, and read-replica? Am I correct in thinking more heap memory is required for the primary servers, whereas more page-cache would be required for read-replicas?
Would we also be able to reduce the amount of memory required on each primary server since the load is distributed? E.g. instead of having 256Gb memory on primary server, have 128Gb on each?
Appreciate any guidance or suggestions, thanks.