There are two cluster configurations:
- PerformOnAnyNode(). This is old. Description: Allow for writes to be forwarded and read requests to be served locally if the current node is not master.
- PreferRandomNode() which was introduced in 3.9.5 version. Description: The client can now be configured to randomly select a node in the cluster instead of preferring master. This can help spread the load, specifically around reads for clients connecting to a cluster.
I have some questions:
- is affected only write operations or r/w?
- is best practices? In which cases I have to use this option? What problems can I have
- is affected only read operations or r/w?
- Can these configurations work separately or together?
- What are the best practices to load balance between cluster nodes?