[ajross]

The database is just a file, so just copy it (rsync works great for compressing this process over a slower network). If you need to replicate without blocking writes, you can do that with some scripting around filesystem (e.g. btrfs) or block layer (LVM) snapshots.

Though again, if you want true streamed (i.e. per query) replication, features like hot standby, etc... that's not part of the model and you're better served by a server-model RDBMS.