Re:Now available: auto-scaling PostgreSQL deployments
by gkelly on Monday January 05, 2015 @06:47PM
(#8842040)
The official PostgreSQL Documentation gives a good start, as well as explaining the many considerations involved in scaling.
http://www.postgresql.org/docs/9.4/interactive/different-rep...
Re:Now available: auto-scaling PostgreSQL deployments
by jeffdavis on Monday January 05, 2015 @07:36PM
(#8842256)
http://www.postgresql.org/docs/9.4/interactive/high-availabi...
I recommend looking at streaming rep + hot stabdby first. Those are built in, officially supported by the project, lower administration effort and fewer pitfalls.
Re:Now available: auto-scaling PostgreSQL deployments
by bmurphy1976 on Monday January 05, 2015 @08:36PM
(#8842449)
Ugh, slony is awful. You use Slony when you have to do a live migration from one version to another (say 9.3 to 9.4) without any downtime. WAL replication is the way to go most of the time. I've used Londiste, it's better than Slony if you need partial replication, but still messy to use compared to WAL.
Personally I haven't used any of the others (i.e. Bucardo).
Re:Now available: auto-scaling PostgreSQL deployments
by akurilin on Tuesday January 06, 2015 @06:01AM
(#8843775)
Would you be so kind as to provide context on why WAL replication is superior to logical replication? I'm very interested in learning more about it.
Also, what specifically did you hate about Slony?
Any resources you'd recommend on this?
Re:Now available: auto-scaling PostgreSQL deployments
by rpedela on Tuesday January 06, 2015 @06:06AM
(#8843792)
WAL replication is currently the official and best supported mechanism for replication. If WAL replication does not suite your particular use case, then that is when to start looking at Slony, Bucardo, etc.
Is there a comprehensive guide on the various options and tradeoffs for this kind of things? Good resources one could use to learn more about it?