The documents highlights possible problems with _SQL_ query intercepts.
I am talking about the actual tuples... i.e. row data rather than the
SQL requests. Please advise if you see any other problems with suggested
approach. Thanks!
Alex.
On Fri, 2007-12-07 at 22:44 -0500, Greg Smith wrote:
> On Fri, 7 Dec 2007, Alex Vinogradovs wrote:
>
> > How about writing a C function (invoked from a trigger) that will send
> > the serialized tuple using say UDP protocol (considering you're syncing
> > on a reliable LAN), and then a simple UDP-listening daemon that will
> > perform the insert into the slave one. If you have multiple slaves, can
> > use that with broadcasting....
>
> I think you are on your way to reinventing "Statement-Based Replication
> Middleware" using triggers and UDP. The great part is that you can look
> into the future and predict the problems you'll run into by reading that
> section of
> http://www.postgresql.org/docs/current/interactive/high-availability.html
> rather then waiting until you run into them yourself.
>
> Seriously, though, not trying to knock you from taking a stab at the
> solution you suggest. Just pointing out that you'll be facing a bit more
> than just a trigger and getting the data moved over the LAN before you're
> done.
>
> --
> * Greg Smith gsmith@gregsmith.com http://www.gregsmith.com Baltimore, MD
>
> ---------------------------(end of broadcast)---------------------------
> TIP 6: explain analyze is your friend