Monitoring the Predictor Resource
You can use performance counters to monitor the performance of the Predictor resource. The Predictor resource has four performance counters, one (the Client Avg Prediction Time counter) that allows you to measure how fast predictions are occurring, and three others that count how often certain operations on the PredictorClient object are being called.
Predictor resource performance counters are only available on computers where the PredictorClient object is used, such as your Web servers. Each counter is aggregated on a per-site basis, so that if you have multiple instances of the PredictorClient object active in a single site, the total number of counters will be totals for the entire site.
The following table lists the performance counters available in the Predictor resource.
| Counter | Description |
| Client Avg Prediction Time | Use this counter to determine the average time it takes for a prediction to be returned. You are provided with an estimate of how much time is being spent each time you call the PredictorClient object. You can also use this counter without inserting timing code on the Active Server Page (ASP) page.
Other uses for this counter are to test how asking for more predictions (by increasing the value for the lMaxPredictions object) affects performance, or to compare the performance of the Predict method on the PredictorClient object for different models. This counter makes measurements in milliseconds. |
| Client Model Loads From File | Use this counter to determine the number of times an analysis model is loaded from a file using the LoadModelFromFile method. |
| Total Client Model Loads from Database | Use this counter to determine the total number of times an analysis model is loaded from a server using the LoadModelFromDB method. |
| Total Client Predictions | Use this counter to determine the total number of client predictions. This counter is useful for keeping track of how often the Predict method on the PredictorClient object is being called. |