Skip to Content

Calibrating queuing network models

Model calibration requires a measured end-to-end response time estimate that can be obtained, for example, from a load testing analysis, and uses it to model the missing resources.

Queuing Network model calibration is used to improve response time estimation in the presence of unmodeled resources, such as network delays, channel contention, or resources that could not be modeled due to lack of data.


Related topics

Creating-and-editing-a-queuing-network-model

Managing-queuing-network-model-scenarios

To apply model calibration

  1. Navigate to Workspace > Domains, Services & Applications > Domain > Works.
  2. In the Works folder, select the queuing network model for which you want to apply calibration. The model details are displayed.
  3. In the Systems table, click Calibration. The Apply calibration page is displayed. 
    QN_model_calibration.png
  4. For each business driver, specify the measure for the Arrival rate and Response time parameters. Providing a measure for a business driver will improve the response time estimate for that customer class and will provide the better identification of unmodeled resources. 
  5. (optional) Set the following parameters:
    • Values expressed in: The measure unit of the arrival rate, for example, Requests per Hour.
    • Calibration type: The type of calibration to perform. Select from the following options:
      • Use a multiplicative factor: This option assumes that the unmodeled resources are uniformly distributed among all systems, thus the response time increase is applied as a multiplicative factor to each system. This is a default option and works well in most situations.
      • Add a delay station: This option adds a delay station (infinite server) to the network. This is the suggested option when the delay is caused by reasons not related to the modeled resources. For example, background network traffic due to other systems in the same LAN or delays due to distance, or high number of network apparatus, such as load balancers, routers, switches, and proxies.
      • Add a queue station: This option adds a queue station to the network. It should be used when there is an evidence that a business driver increase will increase the delay due to a queuing process. A queue station may become the bottleneck of the network.
  6. Click Save. In the model details page, the added calibration resource is displayed in the Systems table.
Warning

Note

If the resource is a Delay or a Queue, you can also perform a what-if analysis with a speed-up factor in the scenario page. For more information, see Managing-queuing-network-model-scenarios.

 

Tip: For faster searching, add an asterisk to the end of your partial query. Example: cert*

BMC Helix Continuous Optimization 22.4