Re: Oracle Parallel Server Option

From: Jonathan Lewis <jonathan_at_jlcomp.demon.co.uk>
Date: Wed, 9 Jun 1999 20:55:10 +0100
Message-ID: <928958324.24911.5.nnrp-07.9e984b29@news.demon.co.uk>

Pete Sharman wrote in message <375D7A12.4264B9D7_at_us.oracle.com>...
>One of the real values in Parallel Server is the scalability beyond the
>capacity of a single node.

A very good point -

A single instance is ultimately limited to a single redo allocation latch and a serial log file write.

I had to design one system which needed to process data so rapidly (and recoverable, and in real-time) that nothing could write a single redo log fast enough to keep up. Running Oracle OPS with 10 instances with strict collision (ping) avoidance was the only way to make it work.

>

Of course, this
>requirement is being negated to a large extent by Cache Fusion in 8i.
>

But Cache Fusion only addresses read/write CR creation, and doesn't address write/write collisions yet. (And the only platform where the performance is significantly better than disk-pinging seems to be NT)

Jonathan Lewis
Yet another Oracle-related web site: www.jlcomp.demon.co.uk Received on Wed Jun 09 1999 - 14:55:10 CDT