Safari Books Online is a digital library providing on-demand subscription access to thousands of learning resources.
6.3.2 Model evaluation and calibration In this section we review each model and ensure that the PATROL - Predict models created from the performance data represent our view of peak periods of resource utilization. That is, we use PATROL - Predict capabilities to validate the reasonableness of the collected performance data for us to use as the base for capacity planning. Open Model Selecting the Open Model icon in the Results pane allows you to work on a selected model (.md) stored on your PC workstation. Evaluate Model After a model is selected, the Evaluate Model option creates a standard set of BMC PATROL for iSeries - Predict reports and graphics. These are located in the Reports and Graphs folders in the navigation pane. Calibrate Model Verify that the values in the model are acceptable compared to measured values from performance data. You can then adjust the predicted 5250-OLTP response times so that it corresponds more closely with the performance tools reports and any unique company collected performance data. The calibration function adds additional delay (representing "other waits") to the 5250-OLTP ("interactive") transaction response time. Non-5250 OLTP transactions can also be calibrated by manually adding delay time to the workload or changing workload properties.