3. PERFORMANCE ANALYSIS TOOLS › 3.2 Relative Importance › 3.2.5 Case Study › 3.2.5.1 Processor Resource Usage

3.2.5.1 Processor Resource Usage

This case study shows how a you can perform a preliminary
analysis of processor usage data and better understand which
general workloads appear to influence the overall system
usage.  This may be thought of as a simple exploratory step
that can help direct other more detailed analyses.

Since performance management is often done in a very
individualized fashion and the actual metrics vary among
sites, it is important to be able to have tools that are
flexible and permit you to choose among a variety of metrics,
measurements, and methods to achieve the best outcome.

DEVELOPMENT OF CONCEPTUAL STUDY

Knowledge gained from this and similar limited analyses can
help you decide which metrics to use in more detailed
studies.  For example, you might decide that simple usage
data, as depicted here, is too limited to be reliable, and
that data from the WLMSECxx or related files would be
required.  But for quick studies and simple "back of the
envelope" analyses, the method used here may be sufficient.

APPLICATION OF CONCEPTUAL STUDY

The systems being examined are SYS1 and SYS2 (these are the
CA MICS SYSID values).  These systems process varied
workloads, which can be loosely grouped together as BATCH,
TSO, STC, and USS.  The systems are used for both system
development and non-scheduled production, and serve a number
of remotely located groups.  The hardware platforms have
changed over time and both LPARs reside on a z-900 based CEC
executing z/OS Version 1.4 or higher.

The DETAIL HARCPU data from a typical day was chosen and the
analyses performed.  The analysis report was reviewed, and
the decision was made to drop the "average" data elements
(CPUAVx) in favor of the "maximum" values (CPUMXx), as stated
above. The query was then re-executed and the results noted.
The next sections explore the results of the report that is
generated from the Relative Importance Analysis developed for
the workload usage study:

     1 - Control Parameters
     2 - Relative Importance Analysis Report