

14. WORKLOAD FORECASTING › 14.1 Functional Description
14.1 Functional Description
Workload Forecasting using Neugents technology produces
nonlinear models that relate different resource consumption
data elements from one or more capacity planning database
files. These forecasts are based on models generated using
Neugents technology, and may use any of the resource element
files produced by CA MICS Capacity Planner. You can use the
same or different files as the source for the Forecast and
Analysis variables in the model. The Forecast element is the
data value being modeled and the Analysis elements are used
to "train" the model before actual forecasting begins.
Upon execution, CA MICS Capacity Planner will read up the
Primary CAP file, select those observations that meet the
filtering criteria specified on panel CAPG700, and sort the
data in preparation for modeling. Likewise, the Analysis
file elements are first selected using panel CAPG710,
observations are selected using the same filtering criteria
used for the Primary CAP file, and validated to ensure that
no missing or zero values exist. If invalid values are
detected, a report is produced and execution is terminated.
If the input data has been determined to be valid, a set of
descriptive statistics of the input values is created and a
report produced accordingly.
The modeling process uses a two-fold approach. First, a set
of data elements is selected from the file specified under
Select Elements from, and the training phase is executed.
The training process enables the model to understand the
relationship(s) between the Analysis element values and both
the actual and predicted values of the Forecast element. Once
the model is trained, a report is generated that contains
statistics used to evaluate the output of the training phase.
These include: the ANOVA R**2 value for the model, produced
by Neugents technology itself; the Model Error value, which
is the average value of the differences between the predicted
and actual values within the forecast data and the Model RMS
Error value, which is the average of the Model Error values,
squared. Examination of these values will help the analyst
determine how effective the model has been in predicting the
Forecast element values.
Once the model has been trained, the forecasting phase is
then executed. In Neugents technology parlance, forecasting
is done by a process called "consulting." Basically, the
trained model is "consulted" internally so that the
relationships established between the Analysis (training)
elements and the Forecast element can be used to determine
future values of the Forecast element. The forecasting phase
is actually executed in two parts. First, the values of the
Analysis elements are extrapolated for the specified number
of cycles. The extrapolation can use either a MANUAL or
AUTOMATIC adjustment option specified by the analyst. If the
MANUAL option is chosen, the analyst supplies an adjustment
percentage for each selected Analysis element. If the
AUTOMATIC option is chosen, CA MICS Capacity Planner examines
the historical data values for each selected Analysis element
and determines an appropriate adjustment factor. In either
case, the adjustment factors are used to generate
extrapolated Analysis element values, and in turn, these are
used to predict new Forecast element values during modeling.
The forecasted values can be optionally saved in a new file,
as can be done with other forecasting methods.
The following reports and printer graphs are automatically
generated by the Workload Forecast Analysis application:
- Model Input Descriptive Statistics
- Model Training Phase Summary
- Model Forecasting Phase Summary report
- Model Forecasting Phase Analysis graph
- Forecast Check report
An optionally requested report called the Model Forecasting
Phase Analysis report can also be produced and used for a
more detailed analysis of the model processing. The graphic
output is produced as simple "printer graphics" for
convenience, and can be produced as either high-resolution
color graphics, or generated as a MQR compatible Tab
delimited file suitable for downloading to a PC workstation
and used for additional analysis and presentation.
Support is available for setting a threshold value alert and
raising a condition if the threshold is exceeded. This
information is shown in the Forecast Check report. The
threshold percentage value you set is first applied against
the baseline forecasted values of the data element chosen to
generate the model. This then sets the upper and lower
threshold limit values. Each observation selected for
processing is compared against these limits. Those
observations found to exceed the limits are flagged as
exceptions and an alert is raised. The alerting process
causes attention to those observations that may impact the
validity or accuracy of the forecast so that the analyst can
take appropriate action. The alerts are summarized on the
MICSLOG and a report is produced detailing all selected
observations and their alert status.
If the ServicePlus Service Desk product is installed and the
MICF global Service Desk option is activated, you may
additionally create a service desk alert. For more
information on using the ServicePlus Service Desk alert
facility, refer to section 4.9 of the CA MICS Resource
Management Planning, Installation, Operation and Maintenance
Guide.
To illustrate the use of Workload Forecasting using Neugents
technology, we have developed a Workload forecasting model
that estimates future processor usage based on workload
consumption. This model relates the historical use of
specified processor workload categories to overall processor
use. A complete description and interpretation of the model
is presented in Section 14.2.
The CA MICS Capacity Planner Option includes features that
utilize CA's U.S. Patented Neugents(R) technology.
In order to exploit this technology the CA MICS Root
Directory under Unix System Services needs to be initialized
and the Neugents technology infrastructure enabled.
For detailed instructions on how to initialize the CA MICS
root directory and enable Neugents technology, review and
complete the steps presented in Chapter 12 of this guide.
Copyright © 2014 CA.
All rights reserved.
 
|
|