Sensitivity analysis of server placement on enterprise network topology through soft computing

We analyze the consequences when both the server placement problem and network topology problem are solved concurrently through soft computing. Both problems are formulated as a combined optimization problem, subject to a set of design and performance constraints while minimizing the enterprise network cost. We have coded the combined optimization problem within a soft computing methodology, which is based on a probabilistic genetic program for automatically searching the design space for good network topologies. The experimental results for synthesizing and optimizing 3-level enterprise network (65 user nodes) for an edge-server placement has demonstrated the effectiveness of our methodology in finding good solutions with a static workload in less than five minutes.