Nonoptimized workload routing

In releases before CICS® TS for z/OS®, Version 4.1, for nonoptimized workload management, and when a coupling facility is not available, workload routing is managed by CICSPlex® SM Workload Manager using a data space owned by a CMAS to share cross-region load and status data.

Every CMAS manages a single WLM data space that it shares with every user CICS region (MAS) that it directly manages. When the CMAS initializes, it verifies and formats the data space with the structures necessary for all workloads associated with the CICS regions that it manages. When the user CICS regions begin routing dynamic traffic, the state of those CICS regions is recorded in this data space. Every 15 seconds, the CICSPLex SM agent in the user CICS region determines the task count at that time and reports to its owning CMAS. The CMAS updates the load count in the target region descriptor of its WLM data space and broadcasts that value to other CMASs participating in workloads associated with the user CICS region.

In an environment in which all user CICS regions are managed by the same CMAS, all routing and target regions refer to the same physical structures in the WLM data space. Dynamic routing decisions are based on the most current load data for a potential routing target region. A routing decision is based on a combination of factors:
  • The number of tasks in the region.
    • Factored as a percentile value, calculated by dividing the target region MAXTASKS setting with the current task count.
  • The health status of the region.
    • Factored by assigning arithmetic weights, depending on whether the region is short-on-storage, taking a transaction dump, taking a system dump, running at its MAXTASKS limit, in a non-responsive state, or in a CPSM stall.
      Note: Health is assessed independently of system availability monitoring (SAM) and is not influenced by SAM settings.
  • The speed of the link between the router and the target.
    • Factored by assigning arithmetic weights, depending on whether the target region is linked to the target by an MRO connection, an XCF connection, an LU6.2 connection, an IPIC connection, or if the target region is the router itself.
      Note: The link speed is not used for link-neutral algorithms.
  • Start of changeThe z/OS WLM health state of the region.
    • Factored by assigning arithmetic weights, depending on the z/OS WLM health indicator (in the range 0 - 100) of a region provided by the z/OS WLM health service. The z/OS WLM health state is factored in only when the z/OS WLM health service is on. A region with a health indicator of zero is assigned a very high penalizing weight, so it is unlikely to receive work if any other region in the same scope has a z/OS WLM health indicator bigger than zero. Regions with indicators in the range 1 - 100 are treated equally in the routing decision.
    End of change
  • If any outstanding CICSPlex SM Realtime Analysis (RTA) Events are associated with the workload.
    • Factored by assigning arithmetic weights depending on the severity of the events outstanding. These events are factored in only when Event name is specified in the WLM specification for the workload or any Transaction Group definitions associated with it.
  • If any transaction affinities are outstanding to override the dynamic routing decision.
    • Regardless of any other factors, if the routing request has an outstanding affinity associated with it, that affinity always overrides the route decision.

Workloads are spread across multiple z/OS images, so additional CMASs are configured to manage the user CICS regions on the disparate LPARs. Each WLM data space must maintain a complete set of structures to describe every CICS region in the workload, not just the CICS regions for which each CMAS is responsible. The WLM data space owned by each CMAS must be periodically synchronized with the WLM data spaces owned by other CMASs that are participating in the workload. This synchronization occurs every 15 seconds from the MASs to their CMASs, and then out to all other CMASs in the workload.

The behavior of CICSPlex SM is different for DTRPGM and DSRTPGM requests:
  • For DTRPGM requests, the routing region calls, from CICS, to decide if the target regions are synchronized with the execution of the request at the selected target, which is then followed by a call back from CICS on completion of the dynamic request. This call allows the router to increment the task load count before informing CICS of the target region system ID, and also to decrement the count on completion of the request.
  • For DSRTPGM requests, the routing region calls, from CICS, to decide if targets are not synchronized with the selected target. Typically, these dynamic requests are asynchronous CICS starts. The router has no notification of when the routed transaction starts or finishes. Because of a lack of notification, CICSPlex SM stipulates that a DSRTPGM target region must also have the workload associated with the region, which transforms the targets into logical routing regions. This call allows the CICSPlex SM routing processes to determine that they are being called at the DSRTPGM target and, therefore, allows the task load count to be adjusted at transaction start and finish.

CICSPlex SM routing regions count the dynamic transaction throughput in a CICSplex, introducing an additional problem in that transactions started locally on the target regions remain unaccountable by the routing regions until a heartbeat occurs. The router transaction counts are not accurately synchronized until two heartbeats have occurred, the first to increment the count, and the second to decrement it again. This discrepancy, however, is not considered as severe as when a router and target are managed by different CMASs.

For multiple CMASs, the router regions evaluate status data for a target region as described in its local WLM data space. If that target region is managed by a different CMAS to that owned by the router, status data describing that target region can be up to 15 seconds old. For DTRPGM requests, this latency does not have a severe effect. For DSRTPGM requests, however, the effect can be significant, particularly when the workload throughput increases. The effect is known as workload batching.

For more information, see Workload batching.



dfha2_eyuaano.html | Timestamp icon Last updated: Sunday, 4 June 2017