stats pack ?/interpretation

[Tilwen]

I have a question about stats pack? I have some db files in the report
that are showing over 20ms read times. In the statspack documentaion
it says that if there are any read times over 20ms for a datafile
something needs to be done about it. However, this database is on a
SAN and I was wondering that the reason some of these datafiles might
have high read times, because those particular datafiles are not in
the SANS cache at the time of the request. Here is an example.
Tablespace Filename
------------------------
----------------------------------------------------
Av Av Av Av Buffer
Av Buf
Reads Reads/s Rd(ms) Blks/Rd Writes Writes/s Waits
Wt(ms)
-------------- ------- ------ ------- ------------ -------- ----------
------
D_CN_OMF /dev/rprod_0016_126
1,009 0 141.3 1.0 285 0 0

This file has only been read 1009 times, so I was thinking that it has
high read times because it gets aged out of the SAN's cache. We are
seeing high I/O wait, about an average of 30% during busy times, and
hardly any CPU usage itself. We have multiple files that have these
high read times. I'm just trying to get some opinions here of other
sites that run on a SAN. We are running oracle 8.1.7.4 IBM SAN, AIX
5L maintenance release3 on a p690 regatta. There are currently 7 cpus
and 8 gig of ram allocated to the node. If you people want, I can
show more of the statspack report. The only performance problems that
I see here are these high average read times.
Mike