Re: Improving speed of copy - Mailing list pgsql-hackers

From Joe Conway
Subject Re: Improving speed of copy
Date
Msg-id 3D8B68BF.1030700@joeconway.com
Whole thread Raw
In response to Re: Improving speed of copy  ("Nigel J. Andrews" <nandrews@investsystems.co.uk>)
List pgsql-hackers
Nigel J. Andrews wrote:
> On Fri, 20 Sep 2002, Shridhar Daithankar wrote:
> 
>>In select test where approx. 15 rows where reported with query on index field, 
>>mysql took 14 sec. and psotgresql took 17.5 sec. Not bad but other issues 
>>eclipse the result..
> 
> I don't know about anyone else but I find this aspect strange. That's 1 second
> (approx.) per row retrieved. That is pretty dire for an index scan. The
> data/index must be very non unique.
> 

Yeah, I'd agree that is strange. Can we see EXPLAIN ANALYZE for that query.

Also, in one of your ealier posts you mentioned a slowdown after raising 
shared buffers from the default 64 to 30000. You might have driven the machine 
into swapping. Maybe try something more like 10000 - 15000.

HTH,

Joe



pgsql-hackers by date:

Previous
From: Kris Jurka
Date:
Subject: Re: Optimizer generates bad plans.
Next
From: Peter Eisentraut
Date:
Subject: Re: [PATCHES] to_char(FM9.9) bug fix