Cross-System Extract record format

The record format of the Cross-System Work Extract Data Set is variable blocked and the block size has to be large enough to contain a performance class record plus the fields CICS PA adds and any other user fields specified. CICS PA will assign default DCB attributes of RECFM=VB,LRECL=8188,BLKSIZE=8192 if they are not specified.

The Cross-System Work Extract that is created is fully compatible with the CICS® Monitoring Facility (CMF) performance data format. However, there are some important differences between the data created by CICS PA and the data collected by CMF. Still, any program that fully exploits the self-defining data format of CMF should have no problem in processing the data created. The important considerations are:
  • Fields
    • Five user fields are in the extract, see How CICS PA creates Cross-System records.
    • Additional user fields are in the extract if requested.
    • All standard CICS CMF fields are in the extract. If a field was missing in the input data, it is set to hexadecimal zeros.
  • Records
    • The records from each network unit-of-work ID are combined into one record.
    • Only performance class records are created.
    • Each SMF (CMF) record created contains only one performance class record.
    • The records are not written in time sequence.
  • IDs and TIME STAMPs
    • The APPLID of the new data is set to MULTIPLE unless overridden by the SYSID operand.
    • The SMF time stamp is set to the latest Stop Time of records in the UOW.
    • The Dictionary START and STOP time stamps are set to the earliest start and latest stop time of records in the UOW.
Two factors make it difficult to create a DSECT for the Cross-System record:
  1. User fields can be added to the record. This adds additional information to the middle of the record, and also adds to data for these fields at the end of the record.
  2. With a maintenance change to CICS PA, the record format can change as long as it remains compatible with the CICS CMF format using the dictionary record supplied at the front of the data set.

The format of the Cross-System Work Extract record is the same as that of a standard CMF performance class record. It corresponds to the default dictionary record for the latest release of CICS. The default is 700. For a complete description of each field and to understand how the fields are collected, see the CICS Performance Guide.

All the CICS fields listed in the table are the “standard” fields included in every data record written to the Cross-System Work Extract data set. In addition, the following five user fields are always written after the CICS fields:

Table 1. Cross-System Work Extract record format: standard user fields
CMF Field ID Length CMF Field Name
CICSPA A001 4 TOTRECS
CICSPA A002 4 APPLRECS
CICSPA A003 4 TRANROUT
CICSPA A004 4 FUNCSHIP
CICSPA A005 4 DPLRECS

Additional user fields can be requested and are placed in the output record following the listed fields. These additional fields cause the variable information in the dictionary to change, and affect the length of the records. The length of each additional field depends on the type of the field (and the specified range for character fields). For each additional user field, there is also an additional halfword inserted. The halfword contains a hex value that increments for each additional field. This increases the offset to each field by 2 for each user field that is requested and increases the size of the record.

The Cross-System Work Extract data set is normally in network unit-of-work ID sequence. Because the records must be sorted by their network unit-of-work, before they are combined, they are not in the same time sequence as when they were created. It is possible to sort the data set by time sequence if required. Simply use any SORT program and sort the time and date in the SMF header. This field is set to the stop time of the data recorded for each data record. To ensure that the dictionary is the first record in a sorted data set, the time and date in its SMF header is set to the earliest start time of any CMF record in the original data.