From: | Jan Wieck <JanWieck(at)Yahoo(dot)com> |
---|---|
To: | Christopher Browne <cbbrowne(at)acm(dot)org> |
Cc: | pgsql-general(at)postgresql(dot)org |
Subject: | Re: PostgreSQL, MySQL, etc., was Re: PostgreSQL is much |
Date: | 2003-12-01 21:09:39 |
Message-ID: | 3FCBAE13.2050306@Yahoo.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-advocacy pgsql-general |
Christopher Browne wrote:
> A long time ago, in a galaxy far, far away, Randolf Richardson <rr(at)8x(dot)ca> wrote:
>> [sNip]
>>>> the difference is that with mysql, nothing pushes the table out of
>>>> memory; it always stays in memory. in postgresql, a big query on
>>>> another tables, or perhaps a vacuum, or other highly active
>>>> applications on the same server can cause the small tables to be
>>>> pushed out of memory. both approches have positives and
>>>> negatives, and in many cases you would probably notice no
>>>> differance
>>>
>>> If this is a small heavily used table, 7.5 with the new ARC buffer
>>> management policy should do much better. Even better, the table
>>> does not actually need to be small: the heavily used portion will
>>> stay in memory where it can be very fast, and the rest will be just
>>> wait its turn on disk.
>>
>> Is this a configurable option by any chance? If not, then
>> perhaps it should be on a per-table, per-index (etc.) basis.
>
> It is a MUCH BETTER thing to have policies that don't require
> configuration effort.
>
> One of the characteristic problems with Oracle is that you have
> immense numbers of "knobs" to tune. You can get it to work "just
> right" if you throw a large enough horde of DBAs at it.
>
> In the case of the ARC policy, what Jan is trying to do is to come up
> with a strategy that is an improvement irrespective of the
> characteristics of the table. If that works out as hoped for, there
> will be no need to "configure" anything in order to take advantage of
> it.
The stuff is in CVS HEAD. Randolf, look at the README file in
src/backend/storage/buffer for some explanations.
Jan
>
> You'd find your applications running faster simply by installing a 7.5
> server; no need to configure anything. It's like getting Pentium chip
> with improved execution strategies; you don't have to recompile
> anything (the way IA-64 mandates it); you just install the app on the
> new box and watch it speed up.
--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck(at)Yahoo(dot)com #
From | Date | Subject | |
---|---|---|---|
Next Message | Ron Mayer | 2003-12-01 21:18:20 | Re: ISO 8601 "Time Intervals" of the "format with time-unit |
Previous Message | Bruce Momjian | 2003-12-01 21:00:10 | Re: BSDCan 2004 |
From | Date | Subject | |
---|---|---|---|
Next Message | Bruce Momjian | 2003-12-01 21:17:29 | Re: What is WAL used for? |
Previous Message | scott.marlowe | 2003-12-01 21:04:33 | Re: Triggers, Stored Procedures, PHP. was: Re: PostgreSQL |