Interpreting the RMF workload activity data

An RMF workload activity report contains snapshot data which is data collected over a relatively short interval. The RMF reports provided in this section are examples of possible data that might be reported for CICS® and IMS in an RMF workload activity report and some possible explanations for the data.

Note: For more information, see z/OS Resource Measurement Facility (RMF) Report Analysis.

The data for a given work request (CICS transaction) in an MRO environment is generally collected for more than one CICS region, which means there can be some apparent inconsistencies between the execution (EXE) phase and the begin-to-end (BTE) data in the RMF reports. This inconsistency is caused by the end of a reporting interval occurring at a point when work has completed in one region but not yet completed in an associated region. Figure 1 illustrates this inconsistency.

For example, an AOR can finish processing transactions, the completion of which are included in the current reporting interval, while the TOR might not complete its processing of the same transactions during the same interval.

Figure 1. Illustration of snapshot principle for RMF reporting intervals
During a particular RMF reporting interval, a transaction TxnA is started in the TOR, and routed to the AOR. Still in that reporting interval, the AOR runs and completes TxnA. TxnA is therefore included in the EXE (execution) total for that reporting interval. Now the RMF reporting interval ends, and a new interval begins. During the new interval, TxnA returns to the TOR, where it is completed. TxnA is therefore included in the BTE (beginning-to-end) total for the new reporting interval.
Figure 2 shows an example of the work manager state section for a service class representing CICS transactions accessing DBCTL.
Figure 2. Hotel reservations service class
REPORT BY: POLICY=HPTSPOL1  WORKLOAD=PRODWKLD SERVICE CLASS=CICSHR RESOURCE GROUP=*NONE PERIOD=1 IMPORTANCE=1
 
-TRANSACTIONS--  TRANSACTION TIME   HHH.MM.SS.TTT
AVG        0.00   ACTUAL                         114 
MPL        0.00   EXECUTION                       78 
 ENDED         216   QUEUED                        36
END/S      0.24  R/S AFFINITY                   0
#SWAPS        0  INELIGIBLE                     0
 EXECTD        216   CONVERSION                     0
AVG ENC   0.00   STD DEV                      270
REM ENC   0.00
MS ENC    0.00
          RESP  --------------------------STATE SAMPLES BREAKDOWN (%) -----------------------  ----STATE------
SUB   P   TIME  --ACTIVE-- READY IDLE  ----------------------WAITING FOR---------------------  SWITCHED SAMPL (%)
TYPE       (%)   SUB  APPL             CONV PROD                                               LOCAL SYSPL  REMOT
CICS BTE  93.4  10.9   0.0   0.0  0.0  89.2  0.0                                                89.2   0.0    0.0
CICS EXE  67.0  19.7   0.0  10.6  0.0  0.0  69.7                                                 0.0   0.0    0.0
The fields in this RMF report describe an example CICS hotel reservations service class (CICSHR). CICS transactions have two phases:
  • The begin-to-end phase (CICS BTE) takes place in the first CICS region to begin processing a transaction. Typically this region is a terminal-owning region (TOR). The TOR is responsible for starting and ending the transaction.
    • The  ENDED  field shows that 216 hotel reservation transactions completed.
    • The  ACTUAL  time shows that the 216 transactions completed in an average transaction time of 0.114 seconds.
  • The execution phase (CICS EXE) can take place in an application-owning region (AOR) and a resource-owning region such as an FOR. In this example, the 216 transactions were routed by a TOR to an AOR.
    • The  EXCTD  field shows that the AORs completed 216 transactions in the interval.
    • The  EXECUTION  time shows that on average it took 0.078 seconds for the AORs to run the 216 transactions. The  EXECUTION  time applies only to the  EXCTD  transactions.

Begin-to-end phase analysis

While running these transactions, CICS records the states the transactions are experiencing. RMF reports the states in the STATE SAMPLES BREAKDOWN (%) section of the report, with one line for the begin-to-end phase, and another for the execution phase. Because there is a CICS BTE and CICS EXE field, you can assume that the time spent in the TOR represents the BTE phase and the time spend in the AOR represents the EXE phase. There is one EXE phase summarizing all the time spend in one or more AORs.

The CICS BTE total field shows that the TORs have information covering 93.4% of the response time, the analysis of which is shown in the remainder of the row. RMF does not have information covering 100% of the 0.114 seconds response time, because it take some time for the system to recognize and assign incoming work to a service class before it can collect information about it.

For most of the 93.4% of the time, the transactions did not run in the TOR, but had been routed locally to an AOR on the same z/OS® image. You can see this by the SWITCHED SAMPL (%) LOCAL field, which is 89.2% of the total state samples. This value accounts for 83.3% of the response time, because 100% of the total state samples correspond to 93.4% of the response time (89.2 x 93.4 / 100 = 83.3%). This value of 89.2% is close, if not equal, to the WAITING FOR CONV field. which indicates that there is no delay in the TOR once the AOR has returned the transactions.

Execution phase analysis

The total execution time is some percentage of the total response time. It is the EXECUTION transaction time (0.078), divided by ACTUAL transaction time (0.114), which is 68.4%. The CICS execution phase (CICS EXE field) covers 67% of the response time. Some of that time the work is active in the AOR, sometimes it is waiting behind another task in the region, but 69.7% of the total state samples in the PROD field (which corresponds to 69.7 x 67 / 100 = 46.7% of the response time) were found outside of CICS, waiting for another product to provide some service to these transactions. Based on the configuration of the system, the transactions are accessing DBCTL.

The LOCL, SYSP, and REMT state percentages appear in the WAITING FOR section if greater than zero and show the percentages of the total state samples the service class was delayed in the these states when CICS was waiting to establish a session. The STATE SWITCHED SAMPL (%) fields LOCL, SYSPL, and REMOT show the percentage of the state samples in which transactions were routed using MRO, MRO/XCF, or z/OS Communications Server connections.