Using tabu search to find optimal switched LAN configurations

Abstract

The performance of a local area network (LAN) can be improved in many cases by the installation of LAN switches. Determining how to partition a LAN and estimating the performance improvements from a given partitioning are difficult problems. An objective function is developed to quantify a partitioning of a LAN based on a tree-structured topology as supported by IEEE 802.1d compliant LAN switches. It is shown that the objective function developed and the use of tabu search metaheuristic results in a method that finds better partitionings and does so faster than existing methods.