we have some postgreswl 9.6.1 servers in two different continent. Latency is from 180 to 250 ms between the two sites.
Actually we use Rubyrep for off-site replication, since servers are completely independent in case of connectivity loss.
We are experiencing some problems with lock tables (rubyrep works with trigger to commit changes to the other server), we think for the uncommon latency.
You might get more mileage from pglogical (https://www.2ndquadrant.com/en/resources/pglogical/), which you can use to ship just the changes, though given my unfamiliarity with Rubyrep maybe that's what it does too (?)
