Closed queueing networks are widely used in many different kinds of scientific and business applications. Since the demands of saving energy and reducing costs are becoming more and more significant with developing technologies, finding a systematic methodology for getting the best arrangement is very important. In this thesis, design rules are proposed for tandem and various other topologies, to help the designer find the best arrangements which maximize the throughput. Our topological arrangements problem (TAP) can be established as: the system has m-service stations in a network and each one may have different design parameters. To relax the queueing system, the original finite buffer queue is decomposed into a buffer and an infinite buffer server system. Mean Value Analysis (MVA) is used to measure the performance of each topology arrangement. Finally, mixed-integer sequential quadratic programming (MISQP) is used to solve the optimization problem and it is compared with enumeration and a simulation model of Arena (a discrete-event model).