Re: Siemens, OPS, Clusters, and 8.0.4....

From: Jonathan Lewis <jonathan_at_jlcomp.demon.co.uk>
Date: Tue, 20 Apr 1999 07:27:52 +0100
Message-ID: <924589788.23091.0.nnrp-07.9e984b29@news.demon.co.uk>

Which Siemens platform ?

The RM1000 was the 'standard' Siemens MPP machine, but that is going non-production in the relativelyshort term.

Blocks are passed (rather than pinged) in 8.1, not in 8.0, so no help from that feature. And I understand that its only Consistent Read blocks that get to the requesting node in 8.1.5, not Current Read - so pinging still occurs.

I have a few problems with partitioned tables on a very large, fast system with lots of partition maintenance.

And in 8.0.3 I found that a lot of global checkpoints were being issued by parallel queries - this may not be the case in 8.0.4 - which for a 1,000 file system (with no readonly tablespaces possible) resulted in a lot of file header updates and wasted time.

Haven't been able to get back and test the effects of 8.0.4 fixes though.

Jonathan Lewis
Yet another Oracle-related web site: www.jlcomp.demon.co.uk Received on Tue Apr 20 1999 - 01:27:52 CDT