Hi John,

The only difference where I am at is that we are talking about AIX.
But there are two database locations on the same file system. Doesn't make much sense from an io spread but that is the way it was set up.
I assume that what you see is because some tables ran into the 2GB limit and in some sort of time crisis, it has been done this way.
IO spreading will not work efficient since locations are on the same file system.
EMC it self has some troubles, but more related to SP's and controllers and how the EMC set up has been done.
Write behind cache could help you, but still, having a table spread on multiple locations, all of them being on the same file system, expect to see quite some IO penalty.
Maybe you could build another file system on a different array and relocate there one of the locations.

Hth,
Armand

John Pan <panjo03@hotmail.com> wrote:
Hi

on the current site I am on - i have run into something I cannot
explain.
The site has multiple database locations for one file. But they are on
the same disk.

I am used to spreading io loading over multiple locations over
multiple disks.
However it does not seem to work on this white and wondering if anyone
else is using simplar environment.

OpemVMS 7.3-1 / EMC disk array / IngresII 2.0.
Could it be the caching characteristics of XFC?
Has anyone else run into similar puzzles.

thanks
_______________________________________________
Info-ingres mailing list



---------------------------------
Do you Yahoo!?
Yahoo! Mail Address AutoComplete - You start. We finish.