The application grouping problem in Software-as-a-Service (SaaS) networks

Software as a Service (SaaS) provided by cloud computing has recently gained widespread adoption. Because of increased competition in the SaaS market, it is essential for a SaaS provider to properly design its computing system. Significant gains can be achieved by efficiently clustering software applications. This paper focuses on the application grouping problem encountered in computer clustering in SaaS networks. We present integer programming formulations and propose an efficient solution procedure based on the column generation technique applied to the problem. The results of a comprehensive computational study show that our column generation-based approach performed very well for large problem instances with optimality gaps varying between 0.00 and 3.02% with an average of 0.98% compared to optimality gaps varying between 0.00 and 230.64% with an average of 99.08% using a standard branch and bound technique as implemented by a state-of-the-art commercial solver.