When transactions take more time to process than expected in your Capacity Planner capacity model, you can review the Simulation Results Latency by Site page to see how long each individual transaction takes to complete, which is the transaction latency. This page shows the time that is expended on various devices involved in the transaction and the total time it takes for a transaction to complete, which is the average latency.
|Transaction latency is measured against the value that you specify for Latency (sec), on the thresholds Update settings page of Simulation Results. For more information, see Simulation Results - Thresholds.|
Replacing slow devices with faster ones can help reduce the amount of time each transaction spends at the device. For example, if you have a modem connection, which is inherently slow, you might want to replace the modem with a faster connection. Alternatively, you might have a CPU or disk that is slow. In either scenario, the devices might not be overloaded but they might process transactions too slowly, indicating the need for faster devices.
High-latency transactions can also be caused by insufficient capacity on a device. In this case, you can use the Capacity Planner Model Editor to replace a saturated device with one or more similar devices that can handle the load of the capacity model.
High-latency transactions might be the indirect effect of software contention. If your capacity model supports locks, you should review their utilization on the Latency by Site page. The steps to address highly-utilized locks are same as high-latency transactions.