|
I've been a Postgres fan since the mid '90s, but I'm gonna have to agree here. There have been tons of Postgres replication systems out there, for a very long time, but in my investigations I never could find one that I liked more than using DRBD and heartbeat. So that's what I did, and it worked well enough. The MySQL replication felt much more integrated and there was one obvious choice and it worked fine for the uses I've had. Did you have to monitor it and sometimes re-replicate it? Yes. Would I have put a replication system in place without monitoring it? No. And these weren't toy systems, one of the environments I used to manage you've probably used if you are in the US (several clients advertised on various superbowls, for example of scale). I always wanted Postgres replication that was easy and well integrated like MySQL. Today I think the replication story in Postgres is much better, but I haven't tried it recently. So is MySQL's though. I LOVE MySQL Galera so far, though I've only used it for low volume stuff. Multi-master, just works, has been super reliable for over a year, but for a very low volume application. The only hitch has been when all nodes go down, you have to manually find the most up to date and start it in bootstrap mode. I wish they could just come up and realize all the nodes are in the cluster and negotiate who is newest automatically. But, it is a small workflow to manually handle it. In any case, either of these replication systems are better than MS-SQL. Not for technical reasons that I know of, it is simply cost concerns. |