HP Process Resource Manager User's Guide

Fine-tuning your PRM configuration
Using prmanalyze to analyze your configuration
Chapter 8168
# prmanalyze -r disk -x dump -1 -t conflict -E -x financials \
-d 4194304 pacct.3
conflict disk report by command name begins at Wed Jul 7 15:27:00 1999
ave Bps disk threshold 4194304.00
unique id ave Bps peak Bps total KB % total
Jul 7 16:20 -
Jul 7 17:00 5.242E+06 7.993E+06 1.229E+07
cp 2.092E+05 5.324E+05 4.903E+05 3.99
mrkt_rsch 1.397E+06 3.734E+06 3.602E+06 29.32
rm 3.056E+05 3.815E+05 7.163E+05 5.83
sales_fcst 4.503E+05 5.404E+05 1.055E+06 8.59
sort 2.841E+05 3.978E+05 6.660E+05 5.42
web_browser 2.142E+06 7.981E+06 5.021E+06 40.86
The mrkt_rsch job was initially timed stand alone on the system, but is
now getting less than one-third, making the report take over three times
as long to print. This delay has the side effect of taking up memory for a
longer period as well.
Clearly, the culprit is the web browser doing file transfers over the net at
an inopportune time. One solution is to give the critical market research
application a disk of its own. However, while waiting for the new disk to
arrive, the administrator can throttle the web_browser application using
disk bandwidth records in PRM. Using the summary report with dump
data filtered out (-x dump), the administrator creates a new PRM
configuration to implement until the new disk arrives. The new
configuration allocates little, if any, bandwidth on the common disks to
the Finance PRM group. This is because the Finance PRM group has its
own private storage.