Re: postgresql performance with multimedia - Mailing list pgsql-performance

From Jan Wieck
Subject Re: postgresql performance with multimedia
Date
Msg-id 412B2DE7.6000809@Yahoo.com
Whole thread Raw
In response to Re: postgresql performance with multimedia  (my ho <mthoatbanjo@yahoo.com>)
Responses Re: postgresql performance with multimedia
List pgsql-performance
On 8/24/2004 1:08 AM, my ho wrote:

> --- Jan Wieck <JanWieck@Yahoo.com> wrote:
>
>> On 8/17/2004 8:44 PM, my thi ho wrote:
>>
>> > Hi,
>> > I am working on a project which explore postgresql
>> to
>> > store multimedia data.
>> > In details,  i am trying to work with the buffer
>> > management part of postgres source code. And try
>> to
>> > improve the performance. I had search on the web
>> but
>> > could not find much usefull information.
>>
>> What version of PostgreSQL are you looking at? Note
>> that the buffer
>> cache replacement strategy was completely changed
>> for version 8.0, which
>> is currently in BETA test. A description of the
>> algorithm can be found
>> in the README file in src/backend/storage/bufmgr.
>
> oki, Thanks for the information. I have a look at 8.0
> beta, but cannot start the statistic collector. (I had
> post this err message before for help, but havent
> really got any clue to fix it)
>> LOG:  could not create IPv6 socket: Address family
> not
>> supported by protocol
>> LOG:  could not bind socket for statistics
> collector:
>> Cannot assign requested address
>> LOG:  disabling statistics collector for lack of
>> working socket

Tom Lane answered to that question. The code in question does resolve
"localhost" with getaddrinfo() and then tries to create and bind a UDP
socket to all returned addresses. For some reason "localhost" on your
system resolves to an address that is not available for bind(2).

>
> btw, what i want to ask here is does postgreSQL have
> any kind of read-ahead buffer implemented? 'cos it
> would be useful in multimedia case when we always scan
> the large table for continous data.

Since there is no mechanism to control that data is stored contiguously
in the tables, what would that be good for?


Jan

--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck@Yahoo.com #

pgsql-performance by date:

Previous
From: Christopher Kings-Lynne
Date:
Subject: Re: Query kills machine.
Next
From: Stef
Date:
Subject: Re: Query kills machine.