How do you convert CPU hours into MIPS for mainframe capacity planning?
Requires Free Membership to View
Although they aren't exact numbers, mainframe capacity planners find CPU hours, MIPS and MSU good indicators for plotting future growth. In addition, reducing MIPS is a huge priority for mainframe managers, because of the weak economy and rising software costs.
Here are some handy shortcuts that can help with learning about CPU, MIPS, mainframe capacity planning and converting CPU hours:
- The number of million instructions per second (MIPS) is generally considered a measure of computing performance; for large servers or mainframes, MIPS is also a way to measure the cost of computing.
- One million service units (MSU) = six MIPS (here MSU and MIPS represent a consumption speed, not an accumulated consumption). So a job that constantly uses 10 MSU (as displayed by monitors), uses 60 MIPS.
- To convert CPU
seconds (accumulated consumption) to MIPS (average consumption speed):
MIPS = (CPU seconds)*EUM/(Elapsed seconds) where EUM=EQUIVALENT UNIPROCESSOR MIPS as defined in the REXX exec.
Example: A job has used 100 CPU seconds during one minute (it is a multi-task job). The system is a 2064-1C5 (1085 MIPS, EUM=217 MIPS). The average consumption speed of this job is 100*217/60=362 MIPS.
More resources on CPU and MIPS:
- DSPRelated.com has a detailed discussion thread that explains MIPS and related issues
- This Google Group weighed the various options for cutting CPU costs, including converting CPU time to MIPS
- Hear one expert’s take on why you need to consider more than MIPS in mainframe capacity planning
- Find out what you need to include in a capacity planning and monitoring program
- Find tips and resources for all aspects of data center capacity planning in this topic center
- Read more Q/As on mainframe CPU utilization
- Get answers to your CPU, MIPS FAQ
Note: This tip is a compilation of advice from various experts on our site.
This was first published in February 2012
Data Center Strategies for the CIO

Join the conversationComment
Share
Comments
Results
Contribute to the conversation