Re: resource allocation ; postgresql-8.0 - Mailing list pgsql-admin

From Scott Marlowe
Subject Re: resource allocation ; postgresql-8.0
Date
Msg-id 1115925192.22291.12.camel@state.g2switchworks.com
Whole thread Raw
In response to Re: resource allocation ; postgresql-8.0  ("Kavan, Dan (IMS)" <KavanD@imsweb.com>)
List pgsql-admin
On Thu, 2005-05-12 at 12:56, Kavan, Dan (IMS) wrote:
> Scott,
> If you had two applications that were each connecting to two different
> databases, have you found that it's beneficial to have to different
> instances of postgres running?

I haven't, but I could see where, if the workloads were very different,
it might be advantageous.

> I have a few reasons why I think it would be better.  One, it would be
> beneficial to customize parameters for each application.  They would be
> different sizes for example and they may each have different amounts of
> people connecting to them.

Possible.  It really depends on how different they are.  There are
benefits to having only one postmaster running, but since you've got
3gigs+ memory, the memory savings probably isn't one of them.

>   Two, from a networking perspective, it may
> be faster ( 2 different ports rather than one )

I doubt that would be a big difference.

> Do you agree with those
> reasons or can you think of other reasons it would be wise?

Well, running two separate clusters under separate postgres superuser
account would afford the maximum in isolation.  Plus you could easily
place each cluster on it's own storage to provide better performance and
keep them from interacting with each other.  If, at some later date you
decide to put them on separate machines you're aleady halfway there.

> It seems
> like it would be more administrative work to have 2 different instances,
> so there may be some cons to that idea as well.

Exactly. You need to schedule separate vacuums for each, etc...

I tend to leave it on one cluster unless performance starts to become an
issue.

pgsql-admin by date:

Previous
From: Tom Lane
Date:
Subject: Re: Query Plan - Index Scan & Seq Scan
Next
From: Tom Lane
Date:
Subject: Re: Compile Error for postgresql-8.0.3