Re: Performance for relative large DB

tobbe wrote:

>Hi Chris.
>
>Thanks for the answer.
>Sorry that i was a bit unclear.
>
>1) We update around 20.000 posts per night.
>
>2) What i meant was that we suspect that the DBMS called PervasiveSQL
>that we are using today is much to small. That's why we're looking for
>alternatives.
>
>Today we base our solution much on using querry-specific tables created
>at night, so instead of doing querrys direct on the "post" table (with
>4-6M rows) at daytime, we have the data pre-aligned in several much
>smaller tables. This is just to make the current DBMS coop with our
>amount of data.
>
>What I am particulary interested in is if we can expect to run all our
>select querrys directly from the "post" table with PostgreSQL.
>
>
20k transactions per day? Doesn't seem too bad. That amounts to how
many transactions per second during peak times? Personally I don't
think it will be a problem, but you might want to clarify what sort of
load you are expecting during its peak time.

>3) How well does postgres work with load balancing environments. Is it
>built-in?
>
>
There is no load balancing "built in." You would need to use Slony-I
and possibly Pg-Pool for that. I don't know about Pg-Pool, but Slony-I
was written in large part by member(s?) of the core development team so
even if it is not "built in" it is not as if it is a team of outsiders
who wrote it.

If you need something proprietary, there are similar solutions with
replication built in which are based on PostgreSQL and licensed under
proprietary licenses.

Best Wishes,
Chris Travers
Metatron Technology Consulting