Data Systems Workload

The Data Systems Workload (DSW) is a non-threadsafe COBOL application that accesses VSAM files. Transactions use Basic Mapping Support (BMS) maps to interface with 3270 terminals. The number of files in use varies depending on configuration, but can be in the range of 16 through 320.

The DSW workload is composed of a number of transactions, where 50% of CICS transactions issue at least one file control (FC) request. On average, six FC requests are issued per CICS task. FC requests are distributed in the following percentages:

69% read
10% read for update
9% update
11% add
1% delete

To simulate users of the application in a controlled manner, IBM Workload Simulator for z/OS is configured to emulate many 3270 terminals. Depending on the configuration, the number of simulated users can be in the range of 1,000 through 4,000.

Several configuration options are available for DSW. Some of these variants are described in the following sections.

DSW static routing

Five CICS regions are configured for the workload. Two terminal-owning regions (TOR) connect to two application-owning regions (AOR). These two AORs then connect to a file-owning region (FOR). Files are accessed in the FOR by using VSAM local shared resources (LSR).

Work enters the system in a TOR and is then transaction-routed to the corresponding AOR. The business logic of the workload then accesses the VSAM data by using CICS function shipping to the FOR. Temporary storage (TS) requests are fulfilled by using local, unrecoverable, auxiliary temporary storage.

All connections use the multi-region operation and cross-memory (MRO/XM) protocol. CICSPlex SM is not used to provide dynamic workload routing in this scenario. Figure 1 shows the topology of DSW in a static routing configuration.

Flow as described in the surrounding text — Figure 1. Topology of the DSW performance workload in static routing configuration

DSW dynamic routing

By using the same business logic and file structure as the static routing variant of the DSW, the application is extended to include the use of VSAM record-level sharing (RLS) and CICSPlex SM dynamic transaction routing. The use of CICSPlex SM introduces the requirement for the following two CICS regions:

CICSPlex SM address space (CMAS) region. The CMAS region is the component of the CICSPlex SM topology responsible for most of the work that is involved in managing and reporting on CICS regions and their resources. Each CICSplex must have at least one CMAS.
Web user interface (WUI) server. The WUI server is a CICS region that acts as a CICSPlex SM application, which uses the API to view and manage objects in the data repositories of CICSPlex SM address spaces.

To remove application affinities and enable dynamic workload distribution, temporary storage requests are fulfilled by using shared temporary storage, which is held in the coupling facility (CF).

As with the static routing configuration, all connections between CICS regions use the MRO/XM protocol.

Transactions enter the system through TORs by using the BMS maps interface, and are then transaction-routed to an AOR. Although the static routing variant that is described in DSW static routing is connected to a single AOR only, all TORs are connected to all AORs in this scenario. A CICSPlex SM workload is defined and installed to route transactions dynamically from the routing regions (the TORs) into the target regions (the AORs). File commands from the business logic use the support for VSAM RLS in CICS TS to access the required data.

Typically, a dynamic routing configuration uses 4 TORs and 30 AORs, although not all regions are highlighted in topology diagrams. Figure 2 shows the topology of the DSW workload when it is configured to use dynamic routing with CICSPlex SM.

DSW dynamic routing by using IPIC

The topology of this workload is the same as described in DSW dynamic routing. The only difference between these variants is that IP interconnectivity (IPIC) is used to facilitate communication between CICS regions, rather than the MRO/XM protocol.

As described Workload design, workloads are designed to minimize any unnecessary overhead or variations in runtime performance. The DSW IPIC workload uses the TCP/IP home address (127.0.0.1) to avoid testing physical networking hardware.