Reset characteristics of statistics counters

When statistics are written to the SMF data set, the counters might be reset.

Reset to zero
Reset to 1
Reset to current values (this applies to peak values)
Other alternative values.

For detailed information about the reset characteristics, see CICS statistics in DSECTs and DFHSTUP report.

The arrival of the end-of-day time, as set by the ENDOFDAY parameters, always causes the current interval to be ended (possibly prematurely) and a new interval to be started. Only end-of-day statistics are collected at the end-of-day time, even if it coincides exactly with the expiry of an interval.

Changing the end-of-day value changes the times at which INTERVAL statistics are recorded immediately. In Figure 1, when the end-of-day is changed from midnight to 1700 just after 1400, the effect is for the interval times to be calculated from the new end-of-day time. Hence the new interval at 1500 as well as for the times after new end-of-day time.

When you change any of the INTERVAL values (and also when CICS is initialized), the length of the current (or first) interval is adjusted so that it expires after an integral number of intervals from the end-of-day time.

These rules are illustrated by the following example. I indicates an interval recording and E indicates an end-of-day recording.

This CICS system is initialized with ENDOFDAY(000000) and INTERVAL(030000). The first interval recording is taken at 0900 hours. At 09.30, the systems administrator changes INTERVAL to 020000. The interval must expire after an integral number of intervals from the end-of-day time, which is currently set to midnight, so to achieve this, an interval recording is taken at 1000 hours. Following the new setting, further interval recordings are taken at 1200 hours and at 1400 hours. At 14.30, though, the systems administrator changes ENDOFDAY to 170000. The interval times are immediately recalculated from the new end-of-day time, and an interval recording is taken at 1500 hours, two hours from the end-of-day time. The end-of-day recording is taken at 1700 hours. The CICS system is left running, so further interval recordings are taken at 1900 hours and 2100 hours. — Figure 1. Resetting statistics counters

If you want your end-of-day recordings to cover 24 hours, set INTERVAL to 240000.

Note: Interval statistics are taken precisely on a minute boundary. Thus users with many CICS regions on a single MVS™ image could have every region writing statistics at the same time, if you have both the same interval and the same end of day period specified. This could cost up to several seconds of the entire CPU. If the cost becomes too noticeable, in terms of user response time around the interval expiry, you should consider staggering the intervals. One way of doing this while still maintaining very close correlation of intervals for all regions is to use a PLT program like the supplied sample DFH$STED which changes the end-of-day, and thus each interval expiry boundary, by a few seconds. See Stagger end-of-day time sample utility program (DFH$STED) for further information about DFH$STED.

Setting STATRCD=OFF reduces the number of times that statistics are written to the SMF data set and the counters are reset to the end-of-day, unsolicited, and requested reset only."