Sequential Simulation Technologies

Ns [13,78] represents the state-of-the-art in networking simulation packages. It supports a variety of models, such as UDP, TCP/IP and wireless. The critical feature of Ns is the ease with which researchers can model networks. This ease of use is largely attributed to its usage of the TCL [80,105] scripting language as its primary interface.

Despite the advances made by Ns in terms of flexibility and ease of use, it is not adequate for simulating large-scale networks. Consider the following: in order for Ns to simulate a simple network with 512 source nodes connected by a duplex link to 512 sink nodes with UDP packets traversing the link configured for drop-tail-queueing, it will allocate almost 180MB of RAM. This simulation processes events at a rate of $\approxeq$1000 per second. For large-scale networks, we require almost 1000 to 100000 times greater performance.

One might believe that because of faster microprocessors, any sequential simulator's performance will adequately scale. That is not the case. In fact, faster microprocessors escalate performance limitations for network simulations. As processors get faster, they are capable of producing more and more packets. Thus, it is absolutely imperative that the networking community invest in the construction of efficient, scalable tools that are not only capable addressing today's modeling problems, but will also continue to improve along with microprocessor technology and enable the time-efficient simulation of even larger capacity networks when the need arises. To accomplish this goal, we believe parallel and distributed simulation techniques hold the key.