To understand performance data, you must first understand the work CICS® performs at your installation. Analyze the work by its basic building blocks: transactions. Group the transactions into categories of similar resource or user requirements and describe each category’s characteristics. Understand the work that CICS performs for each transaction and the volume of transactions expected during any given period. Tivoli® Decision Support for z/OS® can show you various types of data for the transactions processed by CICS.
A service-level agreement for a CICS user group defines commitments in several areas of quantifiable CICS-related resources and services. CICS service commitments can belong to one of these areas:
The following sections describe certain issues and concerns associated with systems management and how you can use the Tivoli Decision Support for z/OS CICS performance feature.
Use the Tivoli Decision Support for z/OS CICS response-time reports to see the CICS application internal response times.
Figure 36 shows the elements of the internal response times.
As described in Decision Support Network Performance Feature Reports, the Network Performance feature generates reports that show the total, end-to-end average response time (operator transit time) for VTAM® applications (for example, a CICS region) by logical unit. The operator transit time consists of the host transit time and the network transit time, which are also shown in the Network Performance feature reports. Using these reports, you can isolate a response-time problem either to the network or to CICS and act on it accordingly. Should the problem be in CICS, you can use the Tivoli Decision Support for z/OS CICS performance feature reports to identify the application causing the response-time degradation.
Poor response time usually indicates inefficient use of either the processor or storage (or both). Tivoli Decision Support-supplied reports can help you isolate a resource as the cause of a CICS performance problem.
If both the Tivoli Decision Support for z/OS CICS performance feature’s statistics component and the Decision Support System Performance feature’s MVS™ component are installed and active, these reports are available for analyzing transaction rates and processor use by CICS region:
Tivoli Decision Support for z/OS produces several reports that can help analyze storage usage. For example, the CICS Dynamic Storage (DSA) Usage report, shows pagepool usage, under the headings 'Pagepool name', 'DSA (bytes)', 'Cushion (bytes)', 'Free storage (bytes)', 'Free storage (pct)', 'Largest free area', 'Getmains', and 'Freemains'.
CICS Dynamic Storage (DSA) Usage
MVS ID ='MV28' CICS ID ='IYCSCTSK'
Date: '2001-01-17' to '2001-01-18'
Free Free Largest
Pagepool DSA Cushion storage storage free
name (bytes) (bytes) (bytes) (pct) area Getmains Freemains
-------- --------- -------- --------- ------- --------- -------- ---------
CDSA 524288 65536 299008 57 245760 2668 2470
ECDSA 5242880 131072 1122304 21 868352 1084154 1067000
ERDSA 11534336 262144 1130496 9 966656 710 16
ESDSA 0 0 0 0 0 0 0
EUDSA 2097152 0 2097152 100 1048576 73620 73620
RDSA 524288 65536 204800 39 122880 40 0
SDSA 262114 65536 249856 95 249856 12 6
UDSA 524288 65536 524288 100 262114 301922 301922
Tivoli Decision Support Report: CICS809
Because CICS Transaction Server for z/OS uses an MVS subtask to page and because an MVS page-in causes an MVS task to halt execution, the number of page-ins is a performance concern. Page-outs are not a concern because page-outs are scheduled to occur during lulls in CICS processing. If you suspect that a performance problem is related to excessive paging, you can use Tivoli Decision Support for z/OS to report on page-ins, using RMF™ data.
The best indicator of a transaction’s performance is its response. For each transaction ID, the CICS transaction performance detail report (in Figure 38) shows the total transaction count and the average response time. The headings are 'Tran ID', 'Tran count', 'Average resp time (sec)', 'Average CPU time (sec)', 'Prog load reqs (avg)', 'FC calls (avg)', 'Exceptions', 'Program storage bytes (max)', 'Getmains < 16MB (avg)', and 'Getmains > 16MB (avg).'.
CICS Transaction Performance, Detail
MVS ID ='MV28' CICS ID ='IYCSCTSK'
Date: '2001-01-17' to '2001-01-18'
Avg Avg Prog Program
resp CPU load Prog FC storage Getmains Getmains
Tran Tran time time reqs loads calls Excep- bytes < 16 MB > 16 MB
ID count (sec) (sec) (avg) (avg) (avg) tions (max) (avg) (avg)
------------ -------- ------- ---- ----- ----- ------ --------- -------- --------
QUIT 7916 0.085 0.017 0 0 18 0 74344 22 0
CRTE 1760 4.847 0.004 0 0 0 0 210176 1 0
AP00 1750 0.184 0.036 0 0 8 0 309800 66 0
PM94 1369 0.086 0.012 0 0 6 0 130096 24 0
VCS1 737 0.073 0.008 2 0 7 0 81200 14 0
PM80 666 1.053 0.155 1 0 62 0 104568 583 0
CESN 618 8.800 0.001 0 0 0 0 41608 0 0
SU01 487 0.441 0.062 4 0 126 0 177536 38 0
...
GC11 1 0.341 0.014 1 0 2 0 37048 10 0
DM08 1 0.028 0.002 0 0 0 0 5040 3 0
======== =========
20359 309800
Tivoli Decision Support Report: CICS101
Use this report to start verifying that you are meeting service-level objectives. First, verify that the values for average response time are acceptable. Then check that the transaction counts do not exceed agreed-to limits. If a transaction is not receiving the appropriate level of service, you must determine the cause of the delay.
For each CICS task, CICS generates an LU6.2 unit-of-work ID. DB2® also creates an LU6.2 unit-of-work ID.
Figure 39 shows how DB2 data can be correlated with CICS performance data using the DB2 token (QWHCTOKN) to identify the task.
If you match the NETUOWPX and NETUOWSX fields in a CICS record to the DB2 token, you can create reports that show the DB2 activity caused by a CICS transaction.
An exception is an event that you should monitor. An exception appears in a report only if it has occurred; reports do not show null counts. A single exception need not be a cause for alarm. An incident is defined as an exception with severity 1, 2, or 3.
The Tivoli Decision Support for z/OS CICS performance feature creates exception records for these incidents and exceptions:
Figure 40 shows an example of an incidents report, giving information on 'Severity', 'Date', 'Time', 'Terminal operator ID', 'User ID', 'Exception ID', and 'Exception description'.
CICS Incidents
DATE: '2001-01-17' to '2001-01-18'
Terminal
operator User Exception Exception
Sev Date Time ID ID ID description
--- ---------- -------- -------- -------- ------------------ ---------------------------
03 2001-01-17 15.42.03 SYSTEM TRANSACTION_ABEND CICS TRANSACTION ABEND AZTS
03 2001-01-18 00.00.00 SYSTEM TRANSACTION_ABEND CICS TRANSACTION ABEND APCT
03 2001-01-18 17.37.28 SYSTEM SHORT_OF_STORAGE CICS SOS IN PAGEPOOL
03 2001-01-18 17.45.03 SYSTEM SHORT_OF_STORAGE CICS SOS IN PAGEPOOL
Tivoli Decision Support report: CICS002
Tivoli Decision Support for z/OS can pass the exceptions to an Information/Management system.
In a CICS multiple region operation (MRO) or intersystem communication (ISC) environment, you can trace a transaction as it migrates from one region (or processor complex) to another and back. The data lets you determine the total resource requirements of the combined transaction as a unit of work, without having to separately analyze the component transactions in each region. The ability to combine the component transactions of an MRO or ISC series makes possible precise resource accounting and chargeback, and capacity and performance analysis.
The CICS UOW Response Times report in Figure 41 shows an example of how Tivoli Decision Support for z/OS presents CICS unit- of-work response times. The headings are 'Adjusted UOW start time', 'Tran ID', 'CICS ID', 'Program name', 'UOW tran count', and 'Response time (sec)'.
CICS UOW Response Times
Time: '09.59.00' to '10.00.00'
Date: 2001-01-18
Adjusted
UOW UOW Response
start Tran CICS Program tran time
time ID ID name count (sec)
-------- ---- -------- -------- ----- --------
09.59.25 OP22 CICSPROD DFHAPRT 2 0.436
OP22 CICSPRDC OEPCPI22
09.59.26 AP63 CICSPRDE APPM00 2 0.045
AP63 CICSPROD DFHAPRT
09.59.26 ARUS CICSPROD DFHAPRT 3 0.158
CSM5 CICSPRDB DFHMIRS
ARUS CICSPRDC AR49000
09.59.27 CSM5 CICSPRDB DFHMIRS 4 0.639
CSM5 CICSPRDB DFHMIRS
MQ01 CICSPROD DFHAPRT
MQ01 CICSPRDD CMQ001
...
Tivoli Decision Support report: CICS902
Users of CICS applications depend on the availability of several types of resources:
In some cases, an application depends on the availability of many resources of the same and of different types, so reporting on availability requires a complex analysis of data from different sources. Tivoli Decision Support for z/OS can help you, because all the data is in one database.
CICS records at the end of each transaction:
This is useful when you require only transaction statistics, rather than the detailed information that CMF produces. In many cases, it may be sufficient to process only this data, since RMF records it as part of its SMF type-72 record. Analysis (and even recording) of SMF records from CMF can then be reserved for those circumstances when the detailed data is needed. Use the MVS Performance Management (MVSPM) component of the System Performance feature of Tivoli Decision Support to report on this data.
When running under goal mode in MVS 5.1.0 and later, CICS performance can be reported in workload groups, service classes, and periods. These are a few examples of Tivoli Decision Support reports for CICS in this environment. Figure 42 shows how service classes were served by other service classes. This report is available only when the MVS system is running in goal mode. The headings are 'Workload group', 'Service class', 'Served class', 'No of times served', 'No of transactions', and 'No of times served per transaction'.
MVSPM Served Service Classes, Overview
Sysplex: 'SYSPLEX1' System: IP02
Date: '2001-01-18' Period: 'PRIME'
Workload Service Served No of times No of No of times
group class class served tx's served per tx
-------- -------- -------- ------------ ------------ --------------
CICS CICSREGS CICS-1 15227 664 22.9
CICS-2 6405 215 29.8
CICS-3 24992 1251 20.0
CICS-4 87155 1501 58.1
CICSTRX 67769 9314 7.3
Tivoli Decision Support report: MVSPM79
Figure 43 shows the average transaction response time trend and how the various transaction states contribute to it.
Note that the times shown for the various transaction states are calculated based on transaction state samples, and so are not necessarily a precise record of the time spent in each state. Adding together the time spent in each of the transaction states (the shaded areas on the graph) gives the average execution time, which is lower than the average response time (the line on the graph). The difference between the response time and the execution time is mainly made up of switch time -- for example, the time the transactions spend being routed to another region for processing.
This report is available when the MVS system is running in goal mode and when the subsystem is CICS or IMS™.
Figure 44 shows how much the various transaction states contribute to the average response time. This report is available when the MVS system is running in goal mode and when the subsystem is CICS or IMS. The report gives information on 'Workload group', 'Service class/Period', 'Ph', 'MVS sys ID', and 'Total state', followed by the percentage of response time spent in each of the states listed in Figure 43.
MVSPM Response Time Breakdown, Overview
Sysplex: 'SYSPLEX1' Subsystem: IP02
Date: '2001-01-18' Period: 'PRIME'
Service MVS Total Activ Ready Idle Lock I/O Conv Distr Local Netw Syspl Timer Other Misc
Workload class sys state state state state wait wait wait wait wait wait wait wait wait wait
group /Period Ph ID (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%)
-------- ---------- --- ---- ----- ----- ----- ----- ----- ----- -- --- ----- ----- ----- ----- ----- ----- -----
CICS CICS-1 /1 BTE CA0 6.6 0.0 0.0 0.0 0.0 0.0 6.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0
C80 29.4 0.0 0.0 0.0 0.0 0.0 14.7 0.0 0.0 0.0 0.0 0.0 14.6 0.0
C90 3.8 0.4 1.3 1.5 0.0 0.2 0.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0
----- ----- ----- ----- ----- ----- ----- --- ----- ----- ----- ----- ----- ----- -----
* 13.3 0.1 0.5 0.5 0.0 0.1 7.2 0.0 0.0 0.0 0.0 0.0 4.9 0.0
/1 EXE CA0 16.0 0.1 0.2 0.1 0.0 15.5 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0
C80 14.9 0.1 0.1 0.1 0.0 3.7 0.0 0.0 0.0 0.0 0.0 0.0 11.0 0.0
C90 14.0 1.6 4.5 4.8 0.0 3.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
----- ----- ----- ----- ----- ----- --- ----- ----- ----- ----- ----- ----- -----
* 14.9 0.6 1.6 1.7 0.0 7.4 0.0 0.0 0.0 0.0 0.0 0.0 3.7 0.0
IMS IMS-1 /1 EXE CA0 20.7 0.4 0.7 0.0 0.0 0.0 19.6 0.0 0.0 0.0 0.0 0.0 0.0 0.0
C80 1.1 0.2 0.1 0.7 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
C90 22.2 5.3 11.9 1.2 0.0 0.2 3.6 0.0 0.0 0.0 0.0 0.0 0.0 0.0
----- ----- ----- ----- ----- ----- ---- ----- ----- ----- ----- ----- ----- -----
* 14.7 2.0 4.2 0.6 0.0 0.1 7.8 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Tivoli Decision Support report: MVSPM73