Advanced replication strategies in MySQL and Oracle mostly cause hair loss and divorce. Everywhere I have worked that uses these databases eventually convinces itself that it needs futuristic replication technology and then goes on to suffer from outages and weird corruption. It turns out, distributed consistency is a hard problem. Oracle and MySQL both benefit monetarily from people assuming it will work and then activating it, only to find out a month or two later that it's a complicated problem that requires actual planning and oversight.

I think Postgres will be a better product when it incorporates similar functionality, but the implementations in MySQL and Oracle lie somewhere on the border between false advertising and reckless endangerment. I appreciate that Postgres not having it as two-lines of configuration sends a message that you are making a commitment. Nearly everything else in Postgres is set-it-and-forget-it. You really want to have a DBA if you're going to try to make replication work. When you read the docs on having a hot standby and how much of a commitment that is and sense yourself getting tired, you're starting to appreciate what you might be getting yourself into with single-master replication, let alone multi-master.