3. PERFORMANCE ANALYSIS TOOLS › 3.2 Relative Importance › 3.2.5 Case Study › 3.2.5.3 Relative Importance Analysis Report

3.2.5.3 Relative Importance Analysis Report

Figure 3-31 shows the Relative Importance Analysis Report for
this case study.

As this report shows, there is no significant difference
between the "average" and "maximum" data element values, and
in both groups, BATCH (CPUxxB) is the least relevant.  The
other workloads are about even in terms of affect on the
CPUTOBTM value.  The BATCH workload value is retained, and
the "maximum" data values are used in the second execution.

                           RELATIVE IMPORTANCE ANALYSIS REPORT

 REPORT FOR: 10/17/03

ELEMENT    MIN. VALUE     MAX. VALUE     MEAN VALUE         STD        CV      REL. IMP.
________ ____________   ____________   ____________   ____________    ______    _________
CPUTOBTM   0:09:53.34     0:53:38.18     0:26:23.91     0:12:40.76     48.03
CPUAVB           3.19          21.80           6.26           2.44     39.05       0.0557
CPUAVE           7.00          13.68          10.13           2.92     28.79       0.1481
CPUAVS         190.21         279.31         236.91          40.26     16.99       0.1456
CPUAVT           2.30          60.70          25.12          16.93     67.39       0.1586
CPUMXB           5.00          25.00           9.57           3.77     39.36       0.0458
CPUMXE           7.00          15.00          10.69           3.36     31.43       0.1398
CPUMXS         195.00         281.00         238.62          40.08     16.80       0.1470
CPUMXT           3.00          63.00          27.20          17.26     63.47       0.1593


TOTAL RELATIVE IMPORTANCE WEIGHTS:                                                 1.0000

 Figure 3-31.  Relative Importance Analysis Report - First Execution

 For the second execution, only the "maximum" (CPUMXx) data           processor times for the workloads would need to be
 values are used.  Not surprisingly, the approximate ratios            considered as well as the simple number of users and a
 of the Relative Importance Index value do not change much             detailed workload characterization study would be required
 between executions.  In a real-life study, the actual                 before using the metrics for any forecasting operation.




                          RELATIVE IMPORTANCE ANALYSIS REPORT

REPORT FOR: 10/17/03


ELEMENT    MIN. VALUE     MAX. VALUE     MEAN VALUE         STD        CV      REL. IMP.
________ ____________   ____________   ____________   ____________   ______    _________
CPUTOBTM   0:09:53.34     0:53:38.18     0:26:23.91     0:12:40.76    48.03
CPUMXB           5.00          25.00           9.57           3.77    39.36       0.1116
CPUMXE           7.00          15.00          10.69           3.36    31.43       0.2493
CPUMXS         195.00         281.00         238.62          40.08    16.80       0.2922
CPUMXT           3.00          63.00          27.20          17.26    63.47       0.3469


TOTAL RELATIVE IMPORTANCE WEIGHTS:                                                1.0000

 Figure 3-32.  Relative Importance Analysis Report - Second Execution