Different psql cluster for graph db?

[akumaigorodski]

This is more of a nice-to-have feature, I wonder what do you think about ability to run two local psql clusters: one for graph stuff and another one for everything else.

The reason for this is streaming replication which can only work for a cluster as a whole. While doing streaming replication I've noticed that graph sync stuff generates the majority of replication traffic while actually being a disposable data which does not need to be backed up. If graph data can be moved to a different cluster then replicating of important node data can be more efficient.