Re: replication

Christopher Browne wrote:
> The world rejoiced as fred(dot)blaise(at)excilan(dot)com (Fred Blaise) wrote:
>
>>Hello
>>
>>I am running postgres 7.4.6 on 2 BSDi boxes. One is live, the other one
>>is a failover. I would like to implement a master-slave replication
>>process.
>>
>>I believe replication has been included in the base package since 7.3.x.
>>
>>Do you have any good pointers/how-to to direct me to?
>
>
> For such cases, I believe that Slony-I is the preferred solution these
> days. The comprehensive documentation may be found in CVS at gborg;
> it may be found online at the URL below.

Hi Christopher,

I've taken a look at Slony-I and found some limitations. Unless I'm
mistaken, Slony-I can't be used to replicate users/groups, and it must
be set up for each individual database. So, for a server where users
and databases are added and removed often, then it seems like Slony-I
may not be the best choice. For simply replicating data in a limited
number of databases, then Slony-I probably performs much better than
pgpool (which has many issues of its own).

We're considering using Slony-I for some applications (where performance
is an issue and user/databases don't change often) and pgpool for others
(where we just need everything replicated).

Michael