7. UNIVARIATE MODEL FORECASTING › 7.6 Case Studies

7.6 Case Studies

To illustrate the use of the univariate modeling techniques,
four case studies are presented in this section.

The first case study, Extended Example Univariate Model
Forecasting, is an extended example that applies Univariate
Model Forecasting to the study of the I/O workload of a
9021-711 and illustrates how to use a geometric moving
average.  This study is a continuation of the study started
in Section 3.3.3.1.

The second study, Performance Group/Period Workload Tracking,
shows how you can use Univariate Model Forecasting to track
the growth of a workload that is defined by a performance
group.  This example shows how to use the data deletion
facility to discard an observation that is unusually low due
to an unexpected event, such as a system outage.  This case
study is a continuation of the study started began in Section
3.3.3.2.

The third study, System Workload and Performance Profile,
looks at trends in CPU utilization and uses the geometric
moving average to dampen the variability in the original
data.  This case study is a continuation of the study started
in Section 3.3.3.5.

The final study, Forecasting New Application Transaction
Values, uses a user-defined transformation to forecast the
new volume of an online workload that you expect to add work
to.  This type of analysis is useful for estimating future
transaction loads, which you can then enter into an analytic
model to measure whether sufficient capacity is available to
meet the demand.

The case studies are presented in the following order:

 1 - Extended Example Univariate Model Forecasting
 2 - Performance Group/Period Workload Tracking
 3 - System Workload and Performance Profile
 4 - Forecasting New Application Transaction Values