* Victoria Parsons <victoria.parsons at streamshield.com> [061201 12:24]:
> Hi,
> 
> I have been using slony to replicate two database from a master machine
> to a varied number of slaves on a production system. The maximum that
> has been tested is 12. I have been asked how many we could get up to.
> >From following the mailing list I have got an idea in my head of no more
> than about 20. This is because of the increased CPU each slon daemon
> uses. I know this could be increased to some extent by getting a more
> powerful machine for my master.

In our experience, slon slaves can put quite a load on the master
postgresql server.

So it's hard to figure out a general recommendation: It depends upon
your hardware, your access patterns, and so on.

> 
> There is talk here of replicating two databases to 1024 machines. I'm
> pretty sure that will fall over in a big heap. Has anyone ever tried
> that many? I have never used the log shipping method - would that help
You could try to replicate from a master to a slave to a slave, but I
guess 1024 is not a workable number this way anyway.

You might also try to do log shipping (slon -a option I think) and
see if that meets your requirements.
> by reducing load on the master? Also I run all slon daemons from the
> master server. Would it become more scalable if I moved the slon daemons
> to each slave in the system.
A little bit.
> 
> If its crazy to try and get that many machines replicating then just
> scream at me now and we'll sort out a different way to send data.
slony is not a messaging solution. Really.

Another thing that you might consider is the fact that a "broken" down
that does not process events need quick reaction:
a) either you bring up the node again
b) or you drop the node from the cluster.

If not you'll start to have non-confirmed events which will make your
replication crawl at best. (In our case, 6 hours is the point where
replication starts to detoriate if one of the slave nodes is down)

Now this is ok with a couple of slaves, but with 2^10 nodes, you'll
almost always have a node that is down.

Andreas