Joseph W. Doherty Mon, 11/17/2008 - 10:42

Insufficient information to answer question.


For instance, a link can be kept at 100% utilization, in theory, by arrival of next packet just as prior packet is transmitted. In this situation, there would be no queuing delay.


Or, if using generalized queuing theory, and assuming the arrival rate is faster than the drain rate, 100% utilization could, in theory, make for an infinite sized queue.


In practice, if the arrival rate is more than the drain rate, queue will drop packets as it overflows. Wait time of packets not dropped, would depend on amount of bits ahead of them and drain rate.


Also in practice, some procotols will slow when they detect drops (and some new ones can slow when they see RTT increase), so it can be very difficult to predict average wait times unless many, many factors are accounted for.

Actions

This Discussion