Interpreting Distribution Appendix output

You can use the detailed output in the Distribution Appendix when you suspect an unusual combination of events was reported in a report summary line. Usually, the average and maximum times or counts are sufficient to highlight a resource usage problem.

However, if you suspect the mean value to be masking an unusual distribution you can draw on the detail contained in the IMS Monitor output records.

For example, suppose you are investigating a change in the scheduling algorithm for a particular transaction and need to know how many transactions were able to be processed for each scheduling of an application program. The following figure shows the processed transactions in a histogram:

Figure 1. Number of transactions processed for each scheduling of an application program

The average is 2.5 transactions per schedule. The distribution in the figure suggests that many schedules were able to process only one or two transactions, and few schedules significantly exhausted the queue. The distribution data for the histogram is as follows:

Number of schedules    1    2    3    4    5    6-10    >10
 
Transactions dequeued  8    7    0    0    2    1       0

The Distribution Appendix presents the histogram data in the form of two lines:

• The first line shows the intervals, prefixed by a cross reference to an individual line on the earlier output.
• The second line gives the number of events occurring in those intervals.

This data appears as follows:

# 950B...0...1...2...3...4...5...10...15...30...90...INF
            8   7   0   0   2   1    0    0    0    0

The cross reference 950B points to a unique report line. For example, the Transaction Queuing report on the appropriate line for the transaction of interest shows 950A,B under the column headed DISTRIBUTION NUMBER. Use the reference number 950B to locate the data in the Distribution Appendix. The 950A reference points to the data for the number of transactions in the queue at schedule time.