11. WORKLOAD CHARACTERIZATION


The Workload Characterization software, provided in the
CA MICS Capacity Planner, enables you to apply a clustering
methodology to the analysis of your installation's workload.
Workload characterization is an attractive approach to
capacity planning and performance management problems because
it allows the number of workload elements that need be
considered in a study to be reduced from tens of thousands to
only a few.

Clustering methods are a statistical extension of scatter
plots to identify similarities and differences between
workloads.  Scatter plots are often difficult to prepare and
depend heavily on visual interpretation of the data.  The
need for visual interpretation limits the use of scatter
plots to two or perhaps three axes.  Clustering overcomes
these disadvantages through its ability to recognize patterns
in multiple dimensions.  Using the CA MICS database as an
input data source for clustering simplifies and extends the
application of the technique.

Note: SAS/STAT is a prerequisite for using Workload
Characterization.

This section contains the following topics:

11.1 Functional Description

11.2 Usage Guidelines

11.3 Standard Output

11.4 Analytic Technique Tutorial

11.5 Component Operation

11.6 Case Studies