It is not possible to lay down any exact rules about whether a file will benefit from conversion to a shared data table. There are many considerations, and an analysis of the potential uses of shared data tables support should be undertaken by someone who understands how the files are used by the various applications and the configuration of the CICS® regions.
The following checklist gives some general guidance. Additional sources of information that could help you to select the files include:
However, the most beneficial input to the selection process is a thorough understanding of the applications and the way in which they use the files.
If your installation is using data tables for the first time, the following checklist gives some general principles to help you select files for defining as data tables.
Information on file usage can be found in the CICS statistics for file control, which is described in File control and File statistics in the CICS Performance Guide . Not all read requests can take advantage of the data table, so you should check the data table information in the CICS statistics report afterwards to verify that the data table is being used effectively. See Interpreting data table statistics for more information.
This section covers just one of the possible inputs to the selection task--the information available from the file statistics.
If you need to share data between more that one MVS™ image, you should investigate using RLS mode to share the files. If, however, your sharing is confined to a single MVS image, you should consider which files have access patterns that make the use of shared data tables beneficial.
Figure 3, Figure 4, and Figure 5 show some extracts from a hypothetical set of file statistics for files accessed in non-RLS mode that are used in the following discussion to demonstrate how CICS statistics can aid the selection process.
The statistics are displayed as they would be reported by the CICS offline formatting utility. Requested file statistics are shown, but Interval or End of Day statistics would be equally suitable. The section of File "Performance Information" statistics, which reports use of VSAM strings and buffers, is not shown here.
The numbers shown in the figures are purely for the purposes of illustration, and you should not expect the statistics at your installation to resemble them. Similarly, the configuration of CICS regions and files has been chosen to highlight certain points; it is not suggested that this is a typical or desirable configuration.
Interpreting data table statistics discusses the statistics reported for files defined as data tables, which you can use to assess the benefits being obtained.
Requested Statistics Report Collection Date-Time 12/25/99-11:51:51 Last Reset 09:00:00 Applid CICFOR Jobname SDTGSTF1
_________________________________________________________________________________________________________________________________
FILES - Resource Information
____________________________
File Dataset Name Dataset RLS DT Time Time Remote Remote Lsrpool
Name Base Dataset Name (If Applicable) Type File Indicator Opened Closed Name Sysid ID
_____________________________________________________________________________________________________________________________
APPLE CIC01.CICOWN.APPLES K NO 07:44:12 OPEN 1
BANANA CIC01.CICOWN.BANANAS K NO 09:45:08 OPEN 1
ORANGE CIC01.CICOWN.CITRUS K NO 10:51:10 OPEN 2
PEAR CIC01.CICOWN.PEARS K NO 07:30:14 OPEN 3
_____________________________________________________________________________________________________________________________
Requested Statistics Report Collection Date-Time 12/25/99-11:51:51 Last Reset 09:00:00 Applid CICFOR Jobname SDTGSTF1
_________________________________________________________________________________________________________________________________
FILES - Requests Information
____________________________
File Get Get Upd Browse Update Add Delete Brws Upd VSAM EXCP Requests RLS req
Name Requests Requests Requests Requests Requests Requests Requests Data Index Timeouts
____________________________________________________________________________________________________
APPLE 2317265 1020 0 1019 21 1 0 11503 310 0
BANANA 536452 1674 20344 1674 908 0 0 2651 70 0
ORANGE 2069454 98560 17831 98327 4543 2563 0 8511 481 0
PEAR 45871 65493 6512 65493 30109 362 0 3773 231 0
____________________________________________________________________________________________________
*TOTALS* 4969042 166747 44687 166513 35581 2926 0 0
Requested Statistics Report Collection Date-Time 12/25/99-11:51:51 Last Reset 09:00:00 Applid CICFOR Jobname SDTGSTF1
_________________________________________________________________________________________________________________________________
FILES - Data Table Requests Information
_______________________________________
File Close Read Recs ¬ Adds from Add Adds rejected Adds rejected Rewrite Delete Highest Storage
Name Type Requests in Table Reads Requests - Exit - Table Full Requests Requests Table Size Alloc(K)
___________________________________________________________________________________________________________________________
DFHST0223 I There are no data table statistics to report.
Requested Statistics Report Collection Date-Time 12/25/99-11:51:38 Last Reset 09:00:00 Applid CICAOR1 Jobname SDTGSTA1
_________________________________________________________________________________________________________________________________
FILES - Resource Information
____________________________
File Dataset Name Dataset RLS DT Time Time Remote Remote Lsrpool
Name Base Dataset Name (If Applicable) Type File Indicator Opened Closed Name Sysid ID
_______________________________________________________________________________________________________________________________
APPLE REMOTE CLOSED CLOSED APPLE CIF1 N
BANANA REMOTE CLOSED CLOSED BANANA CIF1 N
ORANGE REMOTE CLOSED CLOSED ORANGE CIF1 N
ZUCCHINI REMOTE CLOSED CLOSED COURGETT CIA2 N
_______________________________________________________________________________________________________________________________
Requested Statistics Report Collection Date-Time 12/25/99-11:51:38 Last Reset 09:00:00 Applid CICAOR1 Jobname SDTGSTA1
_________________________________________________________________________________________________________________________________
FILES - Requests Information
____________________________
File Get Get Upd Browse Update Add Delete Brws Upd VSAM EXCP Requests RLS req
Name Requests Requests Requests Requests Requests Requests Requests Data Index Timeouts
____________________________________________________________________________________________________
APPLE 1158701 532 0 531 11 1 0 0 0 0
BANANA 305641 0 19067 0 0 0 0 0 0 0
ORANGE 58709 32854 4265 32621 1018 1001 0 0 0 0
ZUCCHINI 78914 0 14765 0 0 0 0 0 0 0
____________________________________________________________________________________________________
*TOTALS* 1601965 33386 38097 33152 1029 1002 0 0
Requested Statistics Report Collection Date-Time 12/25/99-11:51:38 Last Reset 09:00:00 Applid CICAOR1 Jobname SDTGSTA1
_________________________________________________________________________________________________________________________________
FILES - Data Table Requests Information
_______________________________________
File Close Read Recs ¬ Adds from Add Adds rejected Adds rejected Rewrite Delete Highest Storage
Name Type Requests in Table Reads Requests - Exit - Table Full Requests Requests Table Size Alloc(K)
___________________________________________________________________________________________________________________________
DFHST0223 I There are no data table statistics to report.
Requested Statistics Report Collection Date-Time 12/25/99-11:49:31 Last Reset 09:00:00 Applid CICAOR2 Jobname SDTGSTA2
_________________________________________________________________________________________________________________________________
FILES - Resource Information
____________________________
File Dataset Name Dataset RLS DT Time Time Remote Remote Lsrpool
Name Base Dataset Name (If Applicable) Type File Indicator Opened Closed Name Sysid ID
_______________________________________________________________________________________________________________________________
COURGETT CIC02.CICOWN.COURGETT K NO 08:22:15 OPEN 1
LEMON REMOTE NO CLOSED CLOSED ORANGE CIF1 N
_______________________________________________________________________________________________________________________________
Requested Statistics Report Collection Date-Time 12/25/99-11:49:31 Last Reset 09:00:00 Applid CICAOR2 Jobname SDTGSTA2
_________________________________________________________________________________________________________________________________
FILES - Requests Information
____________________________
File Get Get Upd Browse Update Add Delete Brws Upd VSAM EXCP Requests RLS req
Name Requests Requests Requests Requests Requests Requests Requests Data Index Timeouts
____________________________________________________________________________________________________
COURGETT 78914 27469 14765 27469 336472 0 0 8212 481 0
LEMON 2010745 65706 13566 65706 3525 1562 0 0 0 0
____________________________________________________________________________________________________
*TOTALS* 2089659 93175 28331 93175 339997 1562 0 0
Requested Statistics Report Collection Date-Time 12/25/99-11:49:31 Last Reset 09:00:00 Applid CICAOR2 Jobname SDTGSTA2
_________________________________________________________________________________________________________________________________
FILES - Data Table Requests Information
_______________________________________
File Close Read Recs ¬ Adds from Add Adds rejected Adds rejected Rewrite Delete Highest Storage
Name Type Requests in Table Reads Requests - Exit - Table Full Requests Requests Table Size Alloc(K)
___________________________________________________________________________________________________________________________
DFHST0223 I There are no data table statistics to report.
The examples use a hypothetical configuration of three CICS regions. Most of the files used by CICS applications are owned by the file-owning region CICFOR, and the applications mostly run in the application-owning regions CICAOR1 and CICAOR2. This discussion assumes that each of the data sets shown in the statistics reports is a VSAM base KSDS (as indicated by the Dataset Type of K), so any of them can be defined as data tables.
This section focuses on identifying candidates for defining as CICS-maintained data tables, because the decision to define a user-maintained data table is more likely to come from consideration of particular applications than from a study of file performance in general. Because of this focus, none of the statistics shown is for files accessed in RLS mode, because an RLS-mode data set cannot be the source for a CICS-maintained data table.
The statistics also show you which file names in one region are defined to access file names in another region. The Remote Sysid is the name given on the connection between the two regions. In the examples, the SYSID of CICFOR is CIF2 and that of CICAOR2 is CIA2.
The file APPLE is used by applications that run on the application-owning region CICAOR1. It is defined in CICAOR1 as a remote file, and the file definition points to the file APPLE owned by CICFOR. This file would benefit from being redefined in CICFOR as a CICS-maintained data table because it has a high ratio of remote reads (1158701 Get Requests in the time period covered by the reports) to remote updates (11 adds, 1 delete and 531 updates) as seen in Figure 4.
See the CICS Performance Guide for guidance on the meanings of the "FILES - Requests Information" section of a statistics report.
The file BANANA is updated and read on CICFOR, but is also accessed by CICAOR1. Because all the remote accesses are reads and browses, with no updates, the applications running in CICAOR1 would probably see large benefits if BANANA was defined as a data table, and the applications on CICFOR would also benefit by reading from the local data table.
From a study of the statistics in Figure 4 it might appear that ORANGE is not an especially suitable data table candidate, as the numbers of remote retrievals from CICAOR1 (58709 Get Requests and 4265 Browse Requests) are relatively low. However, the remote file LEMON in CICAOR2 also points to ORANGE in CICFOR, so defining ORANGE in CICFOR as a shared CICS-maintained data table would probably benefit the performance of the applications in both AORs.
The file COURGETT owned by CICAOR2 is accessed via the filename ZUCCHINI in CICAOR1. CICAOR1 only reads or browses the file; any updating is issued by the owning region. Also, it is known that these updates are relevant only to the day’s CICS run and do not need to be retained permanently (in fact, they are deleted at shutdown). The file is therefore an excellent candidate for defining as a user-maintained data table. All the updates can then be made to the data table without any VSAM I/O activity, and all the remote retrievals can be made without function shipping.
The file PEAR would probably not benefit much from shared data tables support because it is not accessed remotely and has many update and browse requests. Local browsing does not offer as much benefit as either local reading or any form of remote retrieval, because VSAM browsing (apart from processing of the STARTBR command) is very efficient. This analysis, of course, does not consider the relative importance of the various file accesses; the reading might be done by critical applications, but the time taken for updates might not be important.
The preceding examples illustrate only a small sample of the possible configurations and uses of files that could benefit from shared data tables support.
You could also use shared data tables support to avoid the need to duplicate files or data tables in each region. And, in addition to looking at existing files, you could consider moving files from an AOR to an FOR where this was not practical before because of the cost of file accesses using function shipping.