Message-ID: <45E2A0C1.3010803@archonet.com>
Date: Mon, 26 Feb 2007 08:56:33 +0000
From: Richard Huxton <dev@archonet.com>
User-Agent: IceDove 1.5.0.9 (X11/20061220)
MIME-Version: 1.0
To: "msmbarabino@virgilio.it" <msmbarabino@virgilio.it>
Cc: pgsql-performance@postgresql.org
Subject: Re: Very slow bytea data extraction
References: <110efb42d61.msmbarabino@virgilio.it>
In-Reply-To: <110efb42d61.msmbarabino@virgilio.it>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit

msmbarabino@virgilio.it wrote:
>> If you look at the "actual time" it's completing very quickly indeed. 
> So 
>> - it must be something to do with either:
>> 1. Fetching/formatting 
> the data
>>> 2. Transferring the data to the client.
> 
> I do agree.
> 
>> What 
> happens if you only select half the rows? Does the time to run the 
>> select halve?
> 
> Yes, it does.
> Using pgAdmin, the time to get all 36 
> rows is about 67500ms while it's 24235ms to get only 18 rows.

Hmm - I've seen reports about the traffic-shaping module not being 
install/activated making large data transfers slow. That was on Windows 
2000 though. Might be worth searching the mail archives - I'm afraid I 
run PG on Linux mostly, so can't say for sure.

One other thing I'd test. Make a small table with text columns of the 
same size and see how fast it is to select from that. If it's just as 
slow then it's your network setup. If it's much faster then it's 
something to do with the bytea type.

-- 
   Richard Huxton
   Archonet Ltd