On 3/9/07, Marco Gaiarin <[EMAIL PROTECTED]> wrote:
> Mandi! cedric
> In chel di` si favelave...
>
> > I think, that the answer is true : why your application will attempt to
> > write
> > on a 'slave' node ?
>
> We are rewriting our webapp so they 'read' (SELECT) on a database and
> 'write' (INSERT/UPDATE/DELETE) on another, so we can reed (eventually)
> on slave and write (absolutely) on master, but if there was some
> different way, more low-level/trasparent to the application...
You do understand that slony is an ASYNCHRONOUS replication system,
right? The data you are selecting against on a subscriber will be
slightly behind the data on the origin. You need to keep that in mind
when designing your application or it will cause some very tricky
problems later.
Instead of differentiating based on statements, if you differentiate
based on a higher level abstraction ("reporting" vs. "oltp") you might
have better luck with that kind of a design.
Andrew
_______________________________________________
Slony1-general mailing list
[email protected]
http://gborg.postgresql.org/mailman/listinfo/slony1-general
