Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Mailing Lists -> Oracle-L -> Re: *nix vs MS

Re: *nix vs MS

From: Niall Litchfield <>
Date: Fri, 7 Oct 2005 20:47:24 +0100
Message-ID: <>

On 10/6/05, Robyn <> wrote:
> Everyone,
> We have a fairly large, truly mission critical database (Oracle
> <>) at a remote site that is currently running on
> Microsoft. In
> the past, others have tried to convince mgmt that the system would be
> more reliable on a unix os, but no one has ever been successful in
> obtaining a project to make the change.

It might be instructive to consider the reasons why the project hasn't ever happened previously. It may be that no evidence was presented, it maybe that there are important vendor support or certification reasons, there may be cost issues HP-UX and SUN boxes don't come cheap and so on.

It might also be instructive to consider the motives for suggesting the change. The tone of your post gives the impression not of reasoned argument but of a unshakeable faith in a particular brand of software (or more likely unshakeable doubt in windows). I don't say that this is the case, but appearances and tone can be important.

To my way of thinking, the strongest case for moving this database to
> unix is the track record of this application; it has had far more than
> it's share of issues (bad backups, system crashes, corrupt blocks,
> hung processes, cpu spikes and so on) even though it already gets more
> care and feeding than other databases. (majority of our databases are
> *nix) This is one aspect of what will be presented.

bad backups and corrupt blocks don't make your case for you. They are not a function of the OS. It seems likely that cpu spikes are a application related rather than os related as well, but that should be measurable. That leaves you with system crashes and hung processes. If I were receiving such recommendations I'd want to see something along the lines of how much these events have cost my business in the last year (quarter or 6 months maybe) and a justified suggestion as to by how much these events would be reduced by making the change.

Niall Litchfield
Oracle DBA

Received on Fri Oct 07 2005 - 14:50:49 CDT

Original text of this message