I'm involved with deploying an application that processes a lot of
records each day (millions). The application interacts with an oracle
database which is only as fast as the disks it resides on. Each time I
deploy the product I try my best to explain to our customer system
administrators and disk administrators what the critical filesystems
are and how they should configure their disks.

It's my understanding that there is a way to translate the filesystem
which you see at the o/s level to the disks a particular array is
using; the customer often does not seem to be fully aware of how this
is done. We tend to deploy on systems using emc or hp disk arrays
(1Tb+) and the disk management software is by veritas.

I suppose what I'm asking here is:
(a) is it possible to translate a filesystem to physical disks?
(b) given that (a) is possible what tools are available to monitor and
reduce contention on a disk used by a particular filesystem? what
strategy can you recommend?
(c) when asking for the filesystems to be created what way should the
information be presented to the disk administrators so they know which
filesystems are i/o critical?

-

[ Post a follow-up to this message ]