How to Make Automated Capacity Management Really Work for You
Most companies are asking their capacity planners to find any way to reduce IT costs, typically while managing growing transaction rates, longer program path lengths, increased complexity and heightened end-user expectations for response time.
The task sounds daunting. Haven’t you already tried everything you can to reduce the cost of doing business? If you’ve completed an “essential implementation” of Compuware ThruPut Manager and have created a policy with Service Groups, which describe and control workload processing, you’ve already got the ball rolling toward reduced software costs.
Automated Capacity Management (ACM) is an elective feature of ThruPut Manager that works with IBM’s sub-capacity pricing to help you further reduce your monthly software costs while ensuring that you don’t impact online transaction processing workloads and your most critical batch work. To get the most benefit from the solution, we recommend that you:
- Make sure you’re using sub-capacity pricing on all production LPARS
- Turn on soft-capping on all LPARS; ACM can significantly reduce the risk of implementing soft-capping
- Enable ACM on all LPARs within a CPC (not mandatory, but it works better this way)
But, why do it? If you’ve had experience with soft-capping, you know that it doesn’t differentiate between types of workloads. Once you set it, all workloads will be constrained, which is not a great idea. Ideally, you want to control all aspects of the soft-cap, applying it primarily to work of lower importance.
Since you’ve already built a policy for Service Groups, you can use ACM to specify constraints for these groups. ACM has five capacity levels, each specified as a percentage of the capacity limit. You pick which Service Groups you feel can afford reduced service as each capacity level is attained, as well as tell ACM which constraints to apply.
Let’s look at an example. You’ve set up your levels as follows:
You then give each Service Group or Control Center an ACM constraint. You have some options here:
- Service Class Constraint: Apply limits to jobs currently executing or selected for execution by telling ACM which Capacity Level they should run in
- Maximum Job Constraint: Simply define how many jobs you will let run at the same time
- Job Selection Constraint: Constrain the selection of jobs until they reach a defined threshold, such as between Target and Acceptable
ACM provides easy-to-follow screens that allow you to define thresholds and assign Service Groups to Capacity Levels. You can quickly go in and modify these if you see the need to make changes. Once you’re set up, ACM and ThruPut Manager work to keep your 4HRA as low as possible, keeping your costs down. You can check on the activity in the system log, but you can also use the TMUSER command to see more detail on a specific job.
Use the /SLM Display JESPLEX if you want to see the current 4HRA, capacity level and CPU usage in the last five minutes. It will also show what percent of time the system was capped.
Everyone benefits from the information and results. Managers reap cost benefits, while soft capping is reduced. Online performance won’t be affected. Tech support and performance analysts have less work to do in managing batch, while getting the praise for reducing costs. Operators can readily see what is happening in batch yet will have little effort on their part.
Reducing costs is a critical part of mainframe jobs now. Why not make it easier to succeed by getting ACM up and running?
For more details and examples, check out the Automated Capacity Management Guide under the ThruPut Manager site “Support” tab.
Latest posts by Denise Kalm (see all)
- Breaking Down Dev and Ops Silos with Communication, Collaboration and Trust - April 19, 2018
- 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