| 摘要 |
A method of monitoring the level of performance of a software application running on a network-attached computing device, comprises monitoring information exchange at least one station on the network; measuring at least two performance indicator metrics, such as delay, jitter, loss, response time, throughput, goodput, and object size; and deriving an indicator parameter from a non-linear combination of the indicator metrics. A transformation may be applied to each indicator metric to obtaining a corresponding derived value, and the derived values then additively combined, to obtain the said indicator parameter. The transformation has a first region in which the derived value depends relatively weakly on the corresponding metric, and a second region, in which the derived value depends relatively strongly on the corresponding metric. A score value may be entered by a user, indicative of the user's perception of the performance of the software application, and compared with the derived indicator parameter. The calculation used to derive the indicator parameter may be varied, in dependence on the user-entered score. Diagnostic traces may be triggered in dependence on the derived indicator parameter. Values of the indicator parameter and associated trace data may be collected in a database and collated to diagnose and/or predict problems in the said computer system. |
