CICS Transaction Server for z/OS V3.2

CICS SERVICE MANAGEMENT

A wide range of new functions includes enhancements to the OTE, improvements to systems management, and new capabilities in the CICSPlex SM Web User Interface. There are also improvements in the capacity of applications to deal with large data content, and enhancements to CICS trace and to monitoring and statistics. For more information on the OTE, click here.

Note that the CICSPlex SM TSO End User Interface is removed in CICS TS V3.2.

Online management of program libraries:

A facility is introduced which enables data sets from which program artifacts will be loaded to be defined dynamically without it being necessary to restart the CICS region. This is in addition to the existing means of defining the data sets statically in the DFHRPL concatenation.

Support for Enterprise Workload Manager:

EWLM is the IBM implementation of the Application Response Measurement (ARM) standard from The Open Group. EWLM extends the capabilities of z/OS Workload Management (WLM) services to all members of the IBM eServer family, making possible end-to-end workload monitoring in heterogeneous environments containing multiple, interacting servers.

CICS TS V3.2 provides support for EWLM in a manner transparent to CICS applications, and no change to applications is required to enable them to be monitored with EWLM. CICS support for EWLM means that CICS provides facilities for internal management of ARM correlators, requesting them from EWLM, accepting them from the various transports, forwarding them on, and passing them back to EWLM. By utilizing revised interfaces with WLM on z/OS, CICS effectively indicates to the EWLM Control Center when the processing of a CICS transaction starts and ends, and what kind of transaction it is. CICS flows an ARM correlator with the transaction, so that EWLM is informed of where work is processed and how it flows through the EWLM management domain.

Note: EWLM is only supported for external communications over the following: SOAP, HTTP, the new CICS IP interconnectivity for DPL, IIOP, and MRO. Within a CICS system, correlators are accessible from task related user exits (TRUEs) and certain global user exits (GLUEs) via the performance block token.

Threadsafe file control commands:

In CICS TS V3.2, the commands for accessing local and Record Level Sharing (RLS) VSAM files are now threadsafe. These changes should result in improved performance for threadsafe applications that contain a mixture of commands accessing files and DB2 on an open task control block (TCB). Also, for pure VSAM applications running on an open TCB, there is a higher throughput capability due to better utilization of concurrent CPUs.

The commands that are made threadsafe are READ (including READ UPDATE), REWRITE, DELETE, UNLOCK, WRITE, STARTBR, RESETBR, READNEXT, READPREV, and ENDBR. In addition, the SPI command INQUIRE FILE is now threadsafe.

Note that the commands for accessing files using other access methods (remote files, Shared Data Tables, Coupling Facility Data Tables, and BDAM files) remain non-threadsafe.

Threadsafe definition for system autoinstalled global user exits:

For optimum performance of threadsafe programs, CICS TS V3.2 enables system autoinstalled GLUE programs to be defined as threadsafe.

GLUE programs that are required early during CICS initialization are required to be configured to CICS by using the ENABLE command, and this command can now specify an override of THREADSAFE.

CICS users are encouraged to ensure all GLUE programs are threadsafe, and are defined to CICS using this mechanism.

Threadsafe CICS journaling commands:

CICS threadsafe applications that make use of journaling commands can now benefit from better performance as the following commands are now threadsafe:

• WRITE JOURNALNAME, and WRITE JOURNALNUM

• WAIT JOURNALNAME, and WAIT JOURNALNUM

• XPI WRITE_JOURNAL_DATA

Threadsafe WebSphere MQ commands:

For information on this further threadsafe item, refer to the Components transferred from WebSphere MQ section.

Enhancements to Java configuration and management:

The CICS support for Java programs has been enhanced to provide predictable response times, improved robustness, and better diagnostics. For example:

• It is now possible pre-emptively to start Java Virtual Machines (JVMs) using the PERFORM JVMPOOL command.

• The timeout threshold of JVMs can be changed from the default 30 minutes of inactivity to a period up to 7 days, or indefinitely.

• The termination of JVMs can be actioned against a subset of JVMs within a pool, rather than the whole JVM pool.

• The JVM garbage collection process can now be scheduled to take place within a CICS system program.

• For debugging purposes JVM trace information can now be formatted and integrated with CICS trace facilities.

Note: CICS TS V3.2 does not support Java applications in resettable mode. For more information, refer to the Compatibility section.

Integrated installation of CICSPlex SM:

In CICS TS V3.2, the installation and configuration of the CICSPlex SM components has been simplified by being integrated with the installation of CICS. There is no longer any need to use the EYUISTAR job: CICSPlex SM is installed, along with CICS, using the DFHISTAR job.

In addition, the CICS resource definitions required for CICSPlex SM resources are now dynamically installed, removing the need manually to create definitions in the CICS system definition (CSD). This applies to the CMAS, the Web User Interface server, and the CICSPlex SM agents in the regions managed by CICSPlex SM.

The EYU9XDUT utility can now create the definitions required to start a Web User Interface and its CICSPlex. This removes the former requirement to use the TSO End User Interface or a batch utility to create these definitions.

Enhancements to CICSPlex SM:

Several important functional enhancements are introduced to CICSPlex SM which simplify its use, and provide enhanced data support for system management.

The new utility EYU9XDBT simplifies the setup and use of the CICSPlex SM management environment, making it easier and quicker to exploit the capabilities of CICSPlex SM. It can also be used as an alternative to the batched repository-update facility (BATCHREP) for management and maintenance of data repository resources.

Active task data, which is observable via the TASK resource table and associated views, is now accessible using the TASKRMI base table. The TASKRMI data and application naming fields have also been added to the HTASK record, and a new set of Web User Interface views is shipped to display them. In addition, CICSPlex SM provides full functional support for IP interconnectivity for DPL with the new IPCONN and IPCONDEF base tables. Problem diagnosis in this area is augmented by the new TASKASSC and IPFACIL base tables. CICSPlex SM workload balancing has been upgraded to accommodate dynamic routing over IP connections.

Support for dynamic program library management is provided through the new LIBRARY, LIBDSN, and LIBDEF base tables.

Enhancements to the CICSPlex SM Web User Interface:

The CICSPlex SM Web User Interface is now easier to use, and has a number of new features. The enhancements include:

• Help information is available for all fields. In addition, the terminology used in Web User Interface views and menus is simplified and improved. Care has been taken to ensure that the terminology used is consistent throughout. Long titles used as column headings or button text have been reduced in length, improving screen layout and enabling views to display more data.

• Summary views can now be expanded to display the details of summarized records, through a new link provided to open a new tabular view displaying those records that relate to the selected summary row.

• Map support is added to the Web User Interface, allowing a user to explore the associations between administrative resource definitions within CICSPlex SM in an interactive, diagrammatic manner. This is very useful for understanding the relationships of components within the CICSPlex, and in particular for understanding the topography of applications and associated systems. The Web User Interface map function was also delivered on CICS TS V3.1 in PTF UK15173. (Note that, the CICSPlex SM TSO End User Interface has been removed from CICS TS V3.2).

• The COVC transaction for management of the Web User Interface allows the exporting of items from the Web User Interface repository. The "ALL" option has been added to make it easier to export the whole repository in a single action. The sorting of tabular views has been enhanced by allowing sort on CICS system name.

CICSPlex SM support for new functions:

Operations and systems staff can immediately use CICSPlex SM to manage the new CICS capabilities and resources using its Business Application Services (BAS), operations, and real-time analysis (RTA) functions. These are accessible through the CICSPlex SM Web User Interface and system management API.

Revised CICSPlex SM Web User Interface view and menu packaging:

The manner in which CICSPlex SM Web User Interface views and menus are packaged has been changed, to simplify their import into the Web User Interface repository. Previously, the default set of views and menus were shipped as three members of the SEYUVIEW data set (one for each language); this sometimes resulted in problems in servicing, leading to chains of prerequisite PTFs. Now, there will be three members per map or viewset. Corresponding changes to the COVC Web User Interface management transaction identify an import data set and the members to be imported as an alternative to a transient data queue (TDQ). This change removes the inhibitors to providing timely service to the Web User Interface viewsets and menus.

Customers who have view definitions created on previous releases can import and continue to use these definitions in CICS TS V3.2.

Larger capacity for shared data tables:

A shared data table can now hold more than 2 GB of data, removing the previous limit of a total of 2 GB for all shared data tables owned by a CICS region. This enables CICS to manage large quantities of frequently-accessed application data cached in shared data tables for high performance purposes.

The maximum practical size is now normally limited only by the amount of real storage available. The table control information (consisting of table entry descriptors and index nodes) is moved into two data spaces, so shared data tables no longer use any significant amount of virtual storage in the CICS region. The data table records are stored in multiple data spaces if necessary, with new data spaces being allocated automatically, up to the installation limit for data space storage or a maximum of 100 data spaces (200 GB) per CICS region. There is no limit on the size of an individual shared data table apart from the overall limit for the CICS region.

Support for larger capacity VSAM ESDS files:

CICS TS V3.2 now supports entry sequenced data sets (ESDS) in VSAM greater than 4 GB in length, allowing CICS applications easily to manage larger quantities of data stored in files.

VSAM formerly had the 4-GB limit, but the limit was removed by the introduction of support for an extended format, extended addressing ESDS data set (or "extended ESDS"). This involves the use of a new 64-bit extended relative byte address (XRBA).

CICS writes new format forward recovery log records for extended ESDS files which include the 64-bit RBA. These forward recovery records are supported by CICS VR V4.2 (refer to the CICS Tools information section). In addition, CICS VR V4.2 supports extended ESDS files when the file is offline from CICS and being updated by batch jobs via its batch logging capability.

For the combined environment whereby an extended ESDS file is online to CICS TS V3.2 and is being updated by batch jobs, this is supported by use of the CICS-VSAM RLS function and Transactional VSAM Services (TVS). This requires TVS APAR OA19958.

Improved timing data precision in monitoring and statistics records:

Enhancements are introduced to both the timing precision and the data capacity of CICS monitoring and statistics data. The monitoring clocks now measure elapsed time and CPU time for CICS TCBs to an accuracy of one microsecond, rather than in units of 16 microseconds as hitherto. The previous limit on clock capacity of about 19 hours has also been removed. Previously held as a 32-bit value, it is now a 64-bit value. This means it is only limited by the capacity of the local store clock.

This brings advantages in several areas. With the faster processors available today, the increased precision gives more accurate accounting of CPU time, for charging purposes. It is now easier to determine the exact nature and extent of possible CICS performance problems. In addition, the increase in clock capacity means that it is possible to monitor very long-running CICS transactions.

Note: Clocks for transaction resource class data still use a 32-bit value, measured in units of 16 microseconds.

Compression for monitoring data records:

Data compression can now be performed on the System Management Facilities (SMF) 110 monitoring records produced by the CICS Monitoring Facility. z/OS services are used to compress and expand the data. This should enable greater use of the CICS Monitoring Facility, which can make awareness and analysis of CICS performance problems much easier. It also counteracts the effect of the increased data produced by the improved CICS timer data precision.

Note: The use of compression is optional, and the default is that compression is not used. Therefore tools which process CICS SMF 110 records will not be affected unless compression is switched on. Any tool that is intended to process CICS SMF 110 records for compressed data needs to ensure that it supports the z/OS Data Compression and Expansion Services (CSRCESRV).

Enhancement to CICS statistics:

The monitoring domain global statistics (type MONITOR) are extended to include MVS Workload Manager (WLM) address space goal information. This information will be useful in understanding the MVS WLM settings that are being used for your CICS and CICSPlex SM regions.

Increase in default size of the trace table:

The default size of the trace table is increased from 16 KB to 4 MB. The larger trace table size improves the probability of capturing sufficient information for problem diagnosis should the need arise.

Enhancement allowing DFHTRxxx trace programs to run RMODE(31):

The Generalized Trace Facility (GTF) trace formatting routine calls a suite of CICS trace programs DFHTRxxx. In CICS TS V3.2, these programs have been enhanced to run with RMODE(31) to prevent short-on-storage problems when they are loaded into a CICS region by analysis products, such as IBM Fault Analyzer.

Enhancements to INQUIRE MVSTCB:

Enhancements are made to the INQUIRE MVSTCB command, which shows the addresses and storage information for the MVS TCBs in the CICS address space. These are extensions to include the storage key and amount-in-use of each allocated storage element. The corresponding support is added to CICSPlex SM.

XCF group limit avoidance:

To use the cross-system coupling facility (XCF) to communicate in a sysplex, each participating CICS region joins an XCF group as a member. CICS TS V3.2 now provides the ability to specify the name of the group to join, rather than, as previously, having all the CICS systems in the one DFHIR000 group.

This enables CICS systems to be partitioned into different XCF groups and thereby avoid the XCF limit of only 2047 members being allowed in a single group within a sysplex. For example, production systems could be members of one group, with development and test systems in a separate group.

This capability uses the facility in z/OS for supporting multiple XCF groups, and involves a new system initialization parameter, XCFGROUP, to allow users to specify the XCF group to be joined by the CICS region. In CICSPlex SM, a new CICS-queried attribute, XCFGROUP, is added to the CMAS and CICSRGN base tables.

EXCI has been enhanced to also support XCFGROUP.