> >
> > I find dblink being a nice tool as long as the data volume to
> transfer
> > remains low.
> > I've evaluated it to implement a clustered Postgres environment, but
> > gave it up due to the poor performances.
> > Still waiting for the binary transfer before the next try ;-)
>
> Binary transfer is not a super big deal in terms of performance
> actually in the general case. It's only substantially faster in a few
> cases like timestamp, geo types, and of course bytea. Lack of
> parameterization I find to be a bigger deal actually -- it's more of a
> usability headache than a performance thing.
>
> Also FYI binary dblink between databases is going to be problematic
> for any non built in type unless the type oids are synchronized across
> databases.
>
> merlin
Thanks,
... so I don't really understand where all the time get lost in the
example I posted a few weeks ago:
http://archives.postgresql.org/pgsql-general/2011-09/msg00436.php
Marc Mamin