| 摘要 |
A system for analyzing and estimating average transit delay and high percentiles of delay in a packet network. The system utilizes detailed packet switch measurements that are normally stored in a network database for maintenance and engineering purposes. The system randomly selects source and destination points and samples the stored maintenance measurements for network circuit elements located between the randomly selected source and destination points. These sampled measurements are transmitted to the system processor via electronic mail and the system generates a single server queuing model to determine average and variance of delay for each component. A network-of-queues model is then constructed to determine the end-to-end average delay and end-to-end variance of delay. An Erlang probability distribution is then fit to these end-to-end values to provide statistics of packet transit delay in the packet network.
|
