Hello, PostgreSQL 7.4.8 on Linux here.

We have an ODBC application that must store up to 2660850 rows in a table.

With "store" I mean this logic:

if PK exists then
update columns ;
else
insert new row;
end if ;

I think the best way to do that is to try to update first, then check
out how may rows were updated: if this number is 0, then insert new row.

Is this correct?

How can I speedup this task, apart from issuing a BEGIN TRANSACTION at
the beginning and a COMMIT WORK at the end, of course?

Thank you. Kind regards,

--
Cris Carampa (spamto:cris119@operamail.com)

potevo chiedere come si chiama il vostro cane
il mio è un po' di tempo che si chiama Libero