Most network engineering tools are unsatisfactory. Measurements are not predictive, simulations do not scale, and analysis is limited to oversimplified models. To be more useful, tools should provide robust results, be able to fuse fragmentary results of measurements and simulations, and be pro-active to improve designs. With these objectives in mind, we describe a simplified tool to evaluate the performance of TCP connections and to provide guidelines for network design. Traditional traffic engineering models of communication networks assume that the traffic sources are given. These models are then used to study the quality of service that the network offers to the traffic streams. However, most of the traffic in the Internet is TCP-based. The traffic that a TCP source generates is not specified a priori. Instead, this traffic depends in an essential way on the QoS that the network provides. In this paper, we propose a simple model and we use it to predict the QoS that a network prov...
Jean C. Walrand