Dataset Contention: “No Fighting, No Biting”
As children, many of us read the timeless classic, No Fighting, No Biting, written by Else Holmelund Minarik. It’s a tale of warring alligator siblings and how their mother sought to broker peace between them.
In digging through the treasure chest of ThruPut Manager documentation, I was reminded of this story. There are parallels to the problems of dataset contention. Like the alligators, it’s inevitable that batch jobs will fight over limited system resources.
In a capacity-constrained system, this can really cause delays. Jobs may wait a long time for CPU, memory or I/O channels. Various IBM services, including WLM, serve to balance resource demands by workload priority. But there isn’t much available to broker a resolution when batch jobs fight over a dataset.
Your normal batch schedule has usually been optimized to avoid these issues. However, ad hoc jobs submitted by systems programmers, and even business users, can cause serious “fighting and biting” between batch jobs. This may impact due-out times—unless, of course, you have Compuware ThruPut Manager automating your batch workload.
Dataset Contention Services Enforce the Peace
ThruPut Manager has DCS (Dataset Contention Services) to prevent contention from occurring. When a job is deemed eligible to execute, ThruPut Manager checks the datasets and other conditions to ensure that the job can actually run. If the job fails this verification process, it is rejected from selection and placed on hold.
ThruPut Manager will monitor your system and, when the dataset becomes available, it will automatically release all jobs waiting on this dataset. If a lower priority job has been waiting too long and has been given increased priority, or more commonly, a high priority job is submitted and needs an in-use dataset, ThruPut Manager has the power to take two approaches, depending on what you want it to do for you:
- Nag the user holding the dataset to release the job
- Repossess the dataset from a TSO user
You also have visibility of which job is holding which dataset and what jobs are waiting for one. Parents would find this capability quite useful in ‘toy retention,’ where one child denies having the toy and the other just cries. With ThruPut Manager, you know what is actually happening.
ThruPut Manager does more to eliminate dataset contention issues. By dynamically managing the importance and real-time criticality of a batch job, ThruPut Manager can move a lower priority up the queue when there is a window or when this job becomes the critical path job in a jobstream.
ThruPut Manager runs DCS automatically for you, but you can tailor its services to fit your needs using JAL (Job Action Language) statements. You may not wish to nag certain users or repossess datasets in specific situations. You also have the option to enable ALERTS to notify Operations; they can then decide how to proceed. You choose.
Do you want to ensure that dataset contention is never a performance problem for you? Contact us for more information about the benefits of DCS, another great ‘nugget’ in the ThruPut Manager treasure chest.
Latest posts by Denise Kalm (see all)
- How to Control Your CPU Demand and Costs at the Job Level - November 16, 2017
- Why ThruPut Manager Is the Real-time Cost-cutting Robot for You - November 9, 2017
- Batch Jobs Need More CLASS? Group Workloads by Service Groups - October 24, 2017