6.3: Multipurpose Database workload using IBM MQ
The IBM MQ variant of the MPD workload consists of four queue-owning regions (QORs) listening to a common IBM MQ input queue. Data is received in the QORs using the Distributed Program Link (DPL) CICS-MQ bridge facility. The DPL target program performs a DPL request to another CICS region using the EXEC CICS LINK command. There are eight of these target Application Owning Regions (AORs) which are on the same LPAR as the QORs, connected by using MRO/XM. Work is routed from the QORs to the AORs using CPSM Workload Manager (WLM) operating in queue mode with sysplex-optimized WLM enabled. Programs in the AORs performs various database interactions then return control to the QORs, which send an MQ message to a reply queue.
- 90% of transactions operate in read-only mode, accessing 50 records using EXEC SQL SELECT
- 5% of transactions operate in add/remove mode, adding 10 records using EXEC SQL INSERT and deleting 10 different records using EXEC SQL DELETE
- 5% of transactions operate in update mode, updating 10 records using EXEC SQL UPDATE
- 40% of transactions use a primary key to select records and 60% use a cursor to select records
All CICS regions were running on a single LPAR hosted on an IBM® z17® model ME1, which was configured with eight dedicated CPs for an equivalent machine type of 9175-708. An IBM Storage DS8950F unit was used to provide external storage. The measurement LPAR used z/OS® 3.1. The CICS TS 6.2 and the CICS TS 6.3 environments contained all applicable PTFs issued before 12 January 2026. The workload also used IBM MQ V9.4.3 and IBM Db2 V13.
Table 1 lists the performance results for the MPD using the IBM MQ workload that used CICS TS 6.2.
| External Throughput Rate (ETR) | CICS CPU in QORs (% of a single CP) | CICS CPU in AORs (% of a single CP) | Total CPU per transaction (ms) |
|---|---|---|---|
| 1959.79 | 34.14% | 46.52% | 0.412 |
| 4035.98 | 70.59% | 96.10% | 0.413 |
| 6059.76 | 104.87% | 144.24% | 0.411 |
| 8306.65 | 141.83% | 198.51% | 0.410 |
| 10190.39 | 175.52% | 244.79% | 0.412 |
Table 2 lists the performance results for the MPD using the IBM MQ workload that used CICS TS 6.3.
| External Throughput Rate (ETR) | CICS CPU in QORs (% of a single CP) | CICS CPU in AORs (% of a single CP) | Total CPU per transaction (ms) |
|---|---|---|---|
| 1960.04 | 34.63% | 46.89% | 0.416 |
| 4035.98 | 71.28% | 96.78% | 0.416 |
| 6059.20 | 104.91% | 144.96% | 0.412 |
| 8297.24 | 142.77% | 198.87% | 0.412 |
| 10200.09 | 176.14% | 245.02% | 0.413 |
The figures from Table 1 and Table 2 are plotted in the following graph.
The average CPU per transaction value for CICS TS 6.2 is calculated to be 0.412 ms. The same value for CICS TS 6.3 is calculated to be 0.414 ms. With only a 2 microsecond difference between the two, representing less than 0.5% of the workload cost, and observing the overlapping lines on the graph, the conclusion is that the performance of CICS TS 6.3 is equivalent to that of CICS TS 6.2 for this workload.