Re: Parallel Server and Parallel Query

>>

Has anyone done any Oracle striping on a raw device before? <<

Yes, we run OPS and our entire RDBMS is stripped using 4K blocks which matches the size of our Oracle block. We found that stripping helped distribute the I/O on a dynamic, rapidly growing system.

As far as you question about does Orace know how many disks the data is on when using parallel query? I believe the answer is no, but if extents of the table are in different files you can use parallel query to your advantage. The OS level stripping is invisible to Oracle as Oracle does I/O through the OS. Even when both data sets are spread accross the same disks there can be a benefit to your query.

We do not use parallel query, but have studied buying it. You must be careful though from what I have read and heard. Apply the parallel query option at the sql statement level such as in sql statements that read a large amount of data from one table using a group by then sorts the data for output. One process can read the data and a second can sort it. Trying to use four processes to handle the request would not benefit the query and could result in degrading the query if not the system. To use parallel query you need to set up the tables at creation time so I would limit using it on 1) larger tables and 2) tables most often used in multi-table joins.

Also I would recommend 7.3 for parallel query. 7.1 applied the degree of parallelism at the database level in init.ora while 7.3, I believe, makes you apply it at the query or table level.

Good luck.

Mark Powell -- The only advise that counts is the advise that you follow so follow your own advise Received on Fri Mar 07 1997 - 00:00:00 CST