Re: Oracle on a Beowulf Cluster

From: Billy Verreynne <vslabs_at_onwe.co.za>
Date: Wed, 9 Jun 1999 09:34:48 +0200
Message-ID: <7jl5ak$39h$1@news3.saix.net>

Bart Metzger wrote in message <3756F785.116E21A_at_hotmail.com>...
>Has anyone had experience with this? Is it possible to do
>datawarehousing with this?

You will need Oracle Parallel Server (OPS) to run on a cluster. To my knowledge, OPS has not yet been released for Linux (is it even planned for Linux?).

>Does it make a difference with the kernel, 2.0.x vs 2.2.x. Especially
>as it pertains to the parallel query option.

Wrong question IMHO if you're saying clustering implies parallel query (PQ). Running Oracle on a cluster is not the only way to use PQ. PQ is simply the multithreading of a single query. This can be done on a single CPU box. The only difference is that on a cluster, the multithreading of such a single query can be spread among the nodes of the cluster.

The best place to ask these questions will probably be the Beoweolf e-mail list methink.

Good questions though - I also would like to see OPS on a Linux Beowolf cluster. It should be a better technical solution that a NT cluster.

regards,
Billy Received on Wed Jun 09 1999 - 02:34:48 CDT