How to Automagically Improve Batch Job Performance and Cost
Those who manage batch would love it if someone else took over—or if the batch job would simply disappear. The complexity of batch job streams can feel unmanageable.
Where once you might be able to print out a complete picture of your nighttime work, today you don’t have enough wall space to clearly document it, nor would that be enough. You need to know the critical paths, the required due-out times and more. It’s hard enough to keep up with the shifting flow of online work; most of us can’t get that job done, let alone take on more work.
And yet, it has to be done. Most systems programmers don’t have the luxury of passing the work onto a team of new employees, at least not yet, so they have to get smarter about it. The challenges tend to fall into two buckets:
- Performance: Getting all the batch work done under the SLAs
- Cost: Minimizing the cost of both hardware resources and software
To add to the fun, these two challenges usually have to balance. It can cost more to improve performance, but you might have to sacrifice performance to reduce costs.
What if you could have both? Compuware believes you can and should automate what you don’t have time or resources to do and stop balancing performance against cost.
How to Improve Performance and Reduce Cost
You may not realize that 80 percent of the cost and efficiency of running an application is baked into the code when it’s written. Many tools available for “fixing” poorly performing applications are limited in functionality: buffering, tweaking parameters, moving workloads around, etc. Rarely is there any visualization into the code itself except for hints of inefficiencies, like too many I/Os in a situation where we doubt they are necessary. But you need proof. Developers won’t change code based on your intuition.
Strobe, Compuware’s mainframe application performance management and analysis tool, proves where an application is spending (perhaps wasting) time in any mainframe application, batch or online. The solution can be employed to find inefficiencies in code, especially segments where excessive CPU is being used, and provides a report showing the exact code where the problem resides.
AutoStrobe includes Global Batch Monitoring, which automates the process of detecting performance problems. You don’t even have to set the thresholds; AutoStrobe will look at historical data and set CPU and elapsed time thresholds for you. The result is better performance at lower cost.
With the results, you can go to the developers and demonstrate where they can team with you on saving money and improving performance. It’s a winning play.
Compuware ThruPut Manager
While waiting for the development team to make the recommended changes, you can super-charge batch with ThruPut Manager. This solution adds to the value you’ve gotten from Strobe by optimizing workload initiation, distribution, prioritization and balancing, eliminating dataset contention, SLA management and more.
This all helps with performance and ensures that what is done for batch doesn’t mess up your careful plan for online. And on the cost side, you can use it to optimize your MSU consumption to ensure that your monthly software bills from IBM are minimized.
Document your cost savings and performance improvements and watch the kudos flood in. And you don’t even have to lose sleep over it. Strobe and ThruPut Manager make batch management automagical while you continue to give your company the competitive edge by making online performance hum.
Latest posts by Denise Kalm (see all)
- What Do Performance and Capacity Roles Have to Do with DevOps? - February 1, 2018
- How Developers Can Boost DevOps Through Automated Batch Processing - December 7, 2017
- How to Control Your CPU Demand and Costs at the Job Level - November 16, 2017