You can use the transaction global statistics and transaction
class statistics to see whether the MXT and transaction class limits
are adversely affecting performance.
About this task
To find out how often the MXT and transaction class limits
are reached, look at the transaction global statistics and transaction
class statistics. You can compare the number of times these limits
are reached with the total number of transactions and see whether
the values set for the limits are adversely affecting performance.
Procedure
- To gather statistics relating to the number of times that
the MXT or transaction class limits are reached, you need to use,
at the start of the run, the command CEMT PERFORM STATISTICS RECORD
(or your site replacement) with the keywords TRANSACTION and TRANCLASS.
CEMT PERFORM STATISTICS RECORD [TRANCLASS TRANSACTION]
The statistics are gathered and recorded in the SMF data
set.
- Format this data set by using the statistics utility program,
DFHSTUP. You might find the following DFHSTUP control
parameters useful:
SELECT APPLID=
COLLECTION TYPE=
REQTIME START= ,STOP=
DATE START= ,STOP=
If you correctly code these control
parameters, you avoid formatting information that is unnecessary at
this point. For information about the DFHSTUP utility, see Statistics utility program (DFHSTUP).
Results
If MXT is never reached, or reached only infrequently, it
is not affecting performance. If MXT is reached for 5% of transactions,
this might have a noticeable effect on performance. When the ratio
reaches 10%, there is likely to be a significant effect on performance,
and this could account for some tasks taking a long time to get a
first attach.
What to do next
Consider revising the MXT and transaction class values if
the statistics indicate that they are affecting performance. For guidance
about the performance considerations when you set these limits, see
CICS monitoring facility: Performance and tuning.