September 18th, 2013, 02:15 PM
PostgreSQL database for Big Data
I have application architecture, where using Master/Slave replication strategy. But slaves count about 1100. It's real big. Are there any problems and whether it is at all? Because I don't saw production application with slaves count.
October 8th, 2013, 08:23 AM
The problems will be whether your infrastructure can scale, not really anything about Postgres. You could chain the replication tasks to ease the bandwidth load if you have huge numbers of updates, but again, this is an infrastructure problem, not really a Postgres-specific problem.
October 13th, 2013, 04:33 AM
Theoretically you should be able to setup that many slaves,
but I doubt that it's what you'd want.
You can probably get away with it if the application that use the slaves are read-only. If the 1100 slaves also cause updates then the master will receive a huge amount of updates so it would have to be a gigantic server.
Perhaps the real question is why you want 1100 slaves...