Re: Decide between Postgresql and Mysql (help of

From: "Merlin Moncure" <mmoncure(at)gmail(dot)com>
To: "Jim C(dot) Nasby" <jnasby(at)pervasive(dot)com>
Cc: PFC <lists(at)peufeu(dot)com>, "Scott Marlowe" <smarlowe(at)g2switchworks(dot)com>, Marcos <mjs_ops(at)gmx(dot)net>, pgsql-performance(at)postgresql(dot)org
Subject: Re: Decide between Postgresql and Mysql (help of
Date: 2006-03-29 13:48:22
Message-ID: b42b73150603290548s6894ac6bnecb5c32eb08bae0@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

On 3/28/06, Jim C. Nasby <jnasby(at)pervasive(dot)com> wrote:
> Heh, too quick on the send button...
>
> On Tue, Mar 28, 2006 at 09:42:51PM +0200, PFC wrote:

> Actually, it's entirely possible to do stuff like web counters, you just
> want to do it differently in PostgreSQL. Simply insert into a table
> every time you have a hit, and then roll that data up periodically.
>
> And using MyISAM is no panacea, either. Trying to keep a web counter in
> a MyISAM table means you'll serialize every web page on that counter
> update.

if you want raw speed, use a sequence for a hit-counter. sequences
are wonder-tools and very lightweight. Explain analyze for a sequence
nextval on my desktop box reports 47 microseconds. thats 200k
sequence updates/second. insert into a table (fsync off/cache write,
no keys) is not much slower.

PostgreSQL 8.1 saw a lot of performance improvements...but the most
important (and least publicized) is the reduced latency of simple
queries in high cache enviroments.

merlin

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Merlin Moncure 2006-03-29 14:05:12 Re: Slow performance on Windows .NET and OleDb
Previous Message Gábriel Ákos 2006-03-29 12:22:32 statistics buffer is full