Working with extrapolation models

This topic contains the following sections:

Overview

An extrapolation model estimates metric values as functions of other metrics. Through an initial correlation analysis of existing data, extrapolation estimates the value of a particular metric when the value of another metric changes.

For example, based on a correlation between a server CPU load and a given business driver, data can be extrapolated to determine whether the existing server will be able to sustain a 30% increase of business drivers, and what the average CPU load would be in this condition.

Extrapolation models are of the following types:

  • Extrapolate performance versus load analysis: Correlates performance metrics, such as CPU utilization or response time, with load metrics, such as total events.
  • Extrapolate duration versus load analysis: Correlates a business driver with itself over time (for example, batch elapsed time versus a business driver).
  • Extrapolate load versus load analysis: Correlates a for-instance business driver with a technical business driver, such as daily peak versus total daily business driver.
  • Extrapolate performance versus performance analysis: Correlates system resource metrics.

Example of an extrapolation model for a performance versus load analysis

The region in red represents where the set threshold is exceeded. The number of visits can be increased (to the right end of the red arrow) without exhausting the chosen threshold.

Creating an extrapolation model (With a video)

The following video (4:04) illustrates the process of creating an extrapolation model.

 https://youtu.be/f1mnqbboeY8?list=PLibAMtD70sYEneM-OP7FtO7PrMH-Ce0PL

  1. In the BMC TrueSight Capacity Optimization Console, click Workspace > All Domains> domainName  > Works folder.
  2. In the working area at the right, at the top of the Works table, from the Add list, select Add model.

    In the BMC TrueSight Capacity Optimization Wizard, follow the instructions on each page.
  3. On the Model type page, select EXTRAPOLATION MODEL and click Next.

  4. On the ANALYSIS page, select the analysis that you want to base your extrapolation on. The wizard groups the available analyses by type of Extrapolation model, as described in Working with extrapolation models.

    In the Correlation analysis list, click and select the analysis from the list.

  5. Click Next.
  6. On the Name page, type a Name and Description for the model.
    A default name based on the selected metrics, is displayed. You can change it as required. You can also click  to select the parameters you want to include in the model name.
  7. Click Next.
  8. On the Summary page, click Finish.
    The Extrapolation Model's detail page is displayed, listing the analysis that the model is based on. 

Viewing the details page

The model's detail page displays the entities that the model takes into account.

Extrapolation model: Detail page

  • A link to the analysis that the model is extrapolated from. Clicking on the link takes you to that analysis page.
  • Prediction scenarios table that lists all available scenarios and allows you to add new ones.

Was this page helpful? Yes No Submitting... Thank you

Comments