--- Bart Degryse <Bart.Degryse@indicator.be> wrote:
> Personally I have no good experiences with neither dblink nor dbi-link.
> They perferm only acceptably on very few records.
> My approach now is to use srf plperl functions. That approach requires you
> to have a reasonable knowledge of perl, but is in my experience the fastest
> (or at least slow) one and certainly the most flexible one.
I am sure that the same result can be implemented in pljava as you were able to achieve in plperl
just incase the OP doesn't have a strength in one of the other other pl-languages.
I am not exactly sure how DBLINK works, but I guess that the calling database has to pull all of
(or at least most of) the records from the table of the linked database in order to develop a
JOIN, which would effectivly and always be a SEQ Scan that is pushed through a less that optimal
com connection. If the table was very large, this would probably be a preformance killer.
But if the table is simply a smaller look-up table with less-than a few thousand records that can
easily be materialized in the calling database, DBLINK could be a workable solution.
Regards,
Richard Broersma Jr.