Re: performance of postgresql in replication using slony

pgsql(at)j-davis(dot)com (Jeff Davis) writes:
> On Thu, 2007-07-26 at 01:44 -0700, angga erwina wrote:
>> Hi all,
>> whats the benefits of replication by using slony in
>> postgresql??
>> My office is separate in several difference place..its
>> about hundreds branch office in the difference
>> place..so any one can help me to replicate our dbase
>> by using slony?? and why slony??
>>
>
> This question should be asked on the slony1-general list, you'll get
> more responses there.
>
> The benefit of using slony is that you can read from many servers rather
> than just one.

Indeed.

It would be worth taking a peek at the documentation, notably the
introductory material, as that will give some idea as to whether
Slony-I is suitable at all for the desired purpose.

One thing that "tweaks" my antennae a bit is the mention of having
"hundreds branch office"; there are two things worth mentioning that
would be relevant to that:

- Slony-I is a single-master replication system, *not* a multimaster
system. If someone is expecting to do updates at branch offices, and
that this will propagate everywhere, that is likely not to work out
easily or well.

- If it *is* fair to assess that there will only be one "master",
Slony-I is intended to support a relatively limited number of nodes.
There are no absolute restrictions on numbers of subscribers, but
there are enough communications costs that grow in a polynomial
fashion that I would be quite disinclined to have more than a dozen
nodes in a cluster.
--
(reverse (concatenate 'string "ofni.sesabatadxunil" "@" "enworbbc"))
http://www3.sympatico.ca/cbbrowne/slony.html
HELP! I'm being attacked by a tenured professor!