Web-pages sojourn times in high speed networks

Abstract

Processor sharing (PS) queuing systems have been historically applied for the estimation of file Sojourn times. We present a model for Web-pages sojourn times at the session-layer by incorporating the time spent by TCP in its slow start and congestion avoidance phases. The model is a function of not only the residual capacity of the bottleneck link but also of the packet loss probability and round trip time (RTT) delay. We show that the model effectively captures the Web-pages sojourn times in high speed networks.