DINTV

The DINTV control statement specifies that the Postprocessor is to generate duration reports and indicates the length of the duration interval. The Postprocessor uses the value specified with this control statement also for processing Overview and Exception reports. This allows you to produce trend reports over long time periods.

The duration interval is the length of time each report can cover and should be a multiple of the measurement interval. The syntax of the statement is:

DINTV(hhmm)

where hh denotes the hours and mm denotes the minutes. The maximum is 9960 which is equivalent to 100 hours.

Note:
  1. Before z/OS V1.8 RMF™, exception/overview duration processing was not implemented. If a user specified Postprocessor reports, RMF used the options specified with the REPORTS control statement as a filter to select the SMF records for producing these reports.

    Starting with z/OS V1.8, exception/overview duration processing is supported. If exception/overview duration reporting is used together with normal Postprocessor reports in one job step, the RMF Postprocessor selects all SMF records for producing both the exception/overview reports as well as the normal reports. This leads to uncommon behavior: For example, if a user specifies REPORTS(DEVICE(TAPE)), RMF may additionally produce unwanted DEVICE(DASD) reports.

    Therefore it is not recommended to use REPORTS control statements if you request exception/overview duration reports in the same job.

  2. For duration reports, it is recommended to specify not only the reporting interval but also the date, even if the SMF input data contains only records for those days you want to report on.

The duration interval can be the same length as, or shorter than, the reporting period. If it is shorter, there will be several duration intervals in a reporting period. Figure 1 illustrates how the duration interval relates to the reporting period. Assume a reporting period that covers the twelve hours between 8:00 AM and 8:00 PM over a range of three days. As the figure shows, specifying DINTV(1200) causes the Postprocessor to generate three duration reports, each covering twelve hours of system activity. Specifying DINTV(0600) causes the Postprocessor to generate six duration reports, each covering six hours of system activity. You could also choose a duration interval that does not break the reporting period into equal blocks of time. As the figure shows, specifying DINTV(1000), indicating a duration interval of ten hours, causes the Postprocessor to generate the following reports:

  1. 8:00 AM to 6:00 PM on the first day of the reporting period.
  2. 6:00 PM to 8:00 PM on the first day of the reporting period, and 8:00 AM to 4:00 PM on the second day.
  3. 4:00 PM to 8:00 PM on the second day of the reporting period, and 8:00 AM to 2:00 PM on the third day.
  4. 2:00 PM to 8:00 PM on the third day of the reporting period.

In this case, if you wanted to use the reports to compare system performance over the same hours on each day of the reporting period, it would be difficult because each report covers a different time range and some span two days and include a twelve-hour gap when no reporting was done.

Figure 1. Relationship between Duration Interval and Reporting Period
This figure shows the Relationship between Duration Interval and Reporting Period and is explained in the surrounding text.

The syntax of the control statement allows a maximum duration interval of 99 hours and 60 minutes.

For most effective reporting, the duration interval should relate to the length of the reporting period.

The actual length of time included in the resulting duration report depends on the time within the duration interval when data was actually collected.

Example: Request a duration report for channel path activity that has a duration interval of 12 hours (from 8:00 AM to 8:00 PM) on June 7, 2016:
DATE(06072016,06072016)
RTOD(0800,2000)
DINTV(1200)
REPORTS(CHAN)

However, if channel path activity measurement did not begin until 10:00 AM, and ended at 6:00 PM, the duration report covers the time from 10:00 AM to 6:00 PM The start time of the first measurement interval and the time when the last interval ended appear in the heading of the report.

Midpoint Processing: Postprocessor duration reporting can put data into the incorrect duration interval when the time stamp in the input record differs from the expected time. For example, if your installation uses 15-minutes intervals and specifies RTOD(0900,1000) and DINTV(0100), the time stamp for the 10:00 to 10:15 interval might indicate that the interval began at 09:59:59.997. In this case, the 10:00 record would be reported in the 9:00 to 10:00 duration interval and is one hour and 15 minutes long.

To avoid this problem, the Postprocessor checks whether the midpoint of each interval falls between the duration interval to ensure that a record will not be used unless at least half of the data belongs to the interval.

Therefore, you should specify RTOD(0900,1000) and not RTOD(0859,1000) to get the correct duration report.

This note also applies to Postprocessor interval, exception, and summary reporting.