Workload routing
Workload routing is the directing of transactions or programs among a group of target regions according to the availability, health, and activity levels of those target regions. You can use workload routing in addition to, or in place of, workload separation.
CICSPlex® SM can route a workload among a defined group of target regions by selecting, when each transaction is initiated, the target region that is likely to deliver the best performance.
- Goal
- Queue
- Link neutral goal
- Link neutral queue
If any transaction affinities are outstanding for the transaction being routed, the affinity target region is selected, regardless of the algorithm used.
With the goal and link neutral goal algorithms, the appropriate target region is selected based on the ability of that region to achieve the expected response time. With the queue and link neutral queue algorithms, the appropriate target region is selected such that the load is distributed across a set of target regions.
If the goal or link neutral goal algorithm does not identify a specific target region, the queue or link neutral queue algorithm, respectively, is applied to the remaining set of target regions.
If all the available target regions are equally capable of handling the work, a target region is selected randomly from the group. Therefore, in systems that are lightly loaded, there is no predetermined order in which work is allocated to equally capable target regions.
- Local
- MRO/IRC and MRO/XM (local LPAR)
- MRO/XCF (remote LPAR)
- IPIC (local LPAR)
- IPIC (remote LPAR)
- LU6.2
- Indirect
CICSPlex SM uses the weighting factor of the link as a multiplier against the task load and other factors to determine an overall routing weight. At the end of the evaluation, the region with the lightest weight is usually selected as the target region.
For example, if all other factors are equal, a target region that is connected to its requesting region by using MRO/XCF is preferred to a target region that is connected by using IPIC. IPIC connections between routing and target regions in the same LPAR have a lower weighting than IPIC connections to target regions in a different LPAR, so that if other factors are equal, local IPIC connections are preferred to remote IPIC connections.
Control level for workload routing
To use workload routing, you must specify a default routing algorithm for the workload at the workload specification (WLMSPEC) level. You can optionally specify a routing algorithm at the transaction group (TRANGRP) level. An algorithm specified in a transaction group overrides the default algorithm that is associated with the workload specification.
- QUEUE
- LNQUEUE
- GOAL
- LNGOAL
To change the routing algorithm specified at the workload specification level, you must close down all regions that participate in the workload so that workload is refreshed with the new algorithm specification.
At the transaction group level, you can specify a routing algorithm dynamically. The specified dynamic routing algorithm is applied to every routed dynamic transaction that is associated with the transaction group. Therefore, you can apply an alternative routing algorithm to specific transaction codes in the same workload.
If you specify an alternative routing algorithm at the transaction group level, you can change workload routing characteristics for specific target regions dynamically without stopping your routing region. If you modify an installed transaction group, you must discard its associated WLM definition (WLMDEF) and then reinstall it, so that the transaction group named by the WLM definition is also refreshed. To change the routing algorithm type immediately without discarding and reinstalling the associated WLMDEF, you can use the Active workload transaction groups (WLMATGRP) views and the SET command to change the ALGTYPE attribute.
- INHERIT
- QUEUE
- LNQUEUE
- GOAL
- LNGOAL
Link neutral workload routing
In some situations, link weighting has a strong impact on the routing behavior and can prevent CICS® transaction routing across system boundaries in a single-site parallel sysplex environment. To use workload routing without link weighting, you can use the link neutral goal and link neutral queue algorithms. These algorithms are identical to the goal and queue algorithms, respectively, except that the type of connection between the routing and target region is not considered.
Link neutral algorithms can be useful to route dynamic transactions, for example those that might require services from MVS™ subsystems. With queue and goal algorithms, routers focus dynamic traffic on the systems with the fastest links, which probably reside in the same LPAR. This behavior might overload subsystems in the local MVS image, and under use the remote MVS images that participate in the workload. If you assign these transactions to a transaction group that uses a link neutral algorithm, the routed dynamic traffic is routed more evenly between the local and remote LPARs, and the load on those subsystems is spread.
However, remember that with a link neutral algorithm, if other factors are equal, all target regions have equal preference. The most remote target regions, connected with the slowest telecommunications links, might have equal preference with a locally connected MRO region, or the routing region itself if it is part of the routing target scope. Therefore, consider carefully whether to specify a link neutral algorithm at the workload specification (WLMSPEC) level, because it might affect every dynamically routed transaction. As a result, the workload manager might not select the best target region for dynamic routing traffic and the overall workload throughput might deteriorate.
If you require a link neutral algorithm for a specific transaction set, assign the algorithm to the workload management transaction group (TRANGRP) that identifies those transactions.