Ant colony optimization of virtual machine placement for data latency minimization in cloud systems

In cloud system, the completion time of the slowest processing task is a significant metric when process the data-intensive application by the Map Reduce/Hadoop architectures. Thus an important problem that virtual machine placement for data node problem (VMPDN) is raised. In this paper, considering the successful application of ACO on combinatorial optimization problems, we firstly adopt Max-Min Ant System to solve placement problem and improve the algorithm based on the specific characteristics for VMPDN to accelerate the convergence. We evaluate the performance of our algorithm by giving system simulation. The results demonstrate the effectiveness of the method.