Re: SOLVED: insertion of bytea - Mailing list pgsql-performance

From Chris Mair
Subject Re: SOLVED: insertion of bytea
Date
Msg-id 1130768957.6383.37.camel@dell.1006.org
Whole thread Raw
In response to Re: insertion of bytea  ("Chris Mair" <list@1006.org>)
List pgsql-performance
> I'm CPU-bound with an I/O well below what my disks could do :(
> [...]
>
> PS1: someone off-list suggested using oprofile, which I will do.

I've used oprofile and found out that with my test client (lots of
bytea inserts) the server burns a lot of CPU time in pglz_compress.

I'm using random data and my production data will be closed to random
(due to noise!), so compression is of course pointless.

By using
alter table dbtest alter img set storage external;
I can tell the server not to compress.

On a test box this brought net insert rate up by 50%,
which is enough to meet the requirements.

Thanks again :)

Bye, Chris.





pgsql-performance by date:

Previous
From: Tom Lane
Date:
Subject: Re: performance of implicit join vs. explicit conditions on inet queries?
Next
From: Kelly Burkhart
Date:
Subject: 8.x index insert performance