Re: PostgreSQL Function Language Performance: C vs PL/PGSQL

From: Stephen Frost <sfrost(at)snowman(dot)net>
To: Eliot Gable <egable+pgsql-performance(at)gmail(dot)com>
Cc: pgsql-performance(at)postgresql(dot)org
Subject: Re: PostgreSQL Function Language Performance: C vs PL/PGSQL
Date: 2010-05-26 16:32:51
Message-ID: 20100526163251.GB21875@tamriel.snowman.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

* Eliot Gable (egable+pgsql-performance(at)gmail(dot)com) wrote:
> Thanks for the quick follow-up. So, you are saying that if I can do SPI in
> _PG_init, then I could prepare all my queries there and they would be
> prepared once for the entire function when it is loaded? That would
> certainly achieve what I want. Does anybody know whether I can do SPI in
> _PG_init?

Unless you're using EXECUTE in your pl/pgsql, the queries in your
pl/pgsql function are already getting prepared on the first call of the
function for a given backend connection.. If you're using EXECUTE in
pl/gpsql then your problem might be planning time. Moving that to C
isn't going to change things as much as you might hope if you still have
to plan the query every time you call it..

> The function gets called a lot, but not in the same transaction. It is only
> called once per transaction.

That's not really relevant.. Is it called alot from the same
backend/database connection? If so, and if you're using regular SELECT
statements and the like (not EXECUTE), then they're getting prepared the
first time they're used and that is kept across transactions.

Thanks,

Stephen

In response to

Responses

Browse pgsql-performance by date

  From Date Subject
Next Message Eliot Gable 2010-05-26 16:41:23 Re: PostgreSQL Function Language Performance: C vs PL/PGSQL
Previous Message David Jarvis 2010-05-26 16:30:19 Re: Random Page Cost and Planner