From: | Shane Ambler <pgsql(at)007Marketing(dot)com> |
---|---|
To: | Ray Stell <stellr(at)cns(dot)vt(dot)edu> |
Cc: | Bruno Wolff III <bruno(at)wolff(dot)to>, pgsql-admin(at)postgresql(dot)org |
Subject: | Re: cluster or database? |
Date: | 2007-01-28 17:16:06 |
Message-ID: | 45BCDA56.7000503@007Marketing.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-admin |
Ray Stell wrote:
> On Sat, Jan 27, 2007 at 01:42:52PM -0600, Bruno Wolff III wrote:
>> On Wed, Jan 24, 2007 at 14:32:37 -0500,
>> Ray Stell <stellr(at)cns(dot)vt(dot)edu> wrote:
>>> Is there a drawback to have different clusters on the same
>>> system/diskspace? I have different applications I want to support
>>> initially on the same server/disk array. I was thinking it would be
>>> better/easier to have seperate clusters per app for ease of migration
>>> and security. Is it highly superior just to use different dbs? Why?
>>> Where is this discussed? Thanks.
>> The downside to multiple clusters is that is harder to configure them
>> to play nice with each other in the way of resource utilization because
>> they don't communicate with each other.
>
>
> Thank you for responding. Do you have any documentation that would
> support my efforts to understand some of the issues?
>
You should probably start with
http://www.postgresql.org/docs/8.2/interactive/creating-cluster.html
which will explain more about what a cluster is.
http://www.postgresql.org/docs/8.2/interactive/kernel-resources.html
should give you some insight into the problems system resources can play
with multiple servers on one machine.
There are a couple of ways you can have several clusters on one machine.
1. run initdb on different directories and start PostgreSQL with a
different one (-D option) when you want to work with it (which may fit
in with what you want to do).
2. run multiple copies of PostgreSQL each with it's own cluster, port
and using more system resources.
2 is where the resource problems come in - each server will need a
number of resources which is limited by system settings and often will
need a recompiled kernel to allow higher limits.
You shouldn't have any problems developing multiple databases on the one
server/cluster - each db is entirely separate from the others and you
would only have problems if you are customizing the system catalog and
want differences with each db you are working on.
When you come to install your db on the client server you create a new
cluster for their server and create the users/groups they will use then
create the database for them with the structure you want in it. You only
need to transfer the one database structure from your development server
not your entire cluster.
--
Shane Ambler
pgSQL(at)007Marketing(dot)com
Get Sheeky @ http://Sheeky.Biz
From | Date | Subject | |
---|---|---|---|
Next Message | Peter Eisentraut | 2007-01-28 17:21:50 | Re: [ADMIN] server process (PID xxx) was terminated by signal |
Previous Message | Tom Lane | 2007-01-28 15:00:08 | Re: [ADMIN] server process (PID xxx) was terminated by signal |