Heuristics for Optimal Placement and Migration of Virtual Machines

Virtualization is widely used owing to its advantages, such as flexibility, scalability, and cost reduction. One important advantage is the decrease in power consumption, which is obtained by concentrating virtual machines (VMs) into a fewer physical machines (PMs). This is done by optimally placing VMs to their hosts. This placement problem is an intractable combinatorial optimization problem. The optimal placement will also change if the load on the VMs changes with time. This change necessitates the migrations of VMs among PMs. The number of executed migrations should be small because migrations offer load on the network. Thus, both power consumption and number of migrations should be minimized. This research examines algorithms that solve this optimization problem. The examined algorithms include two metaheuristics: simulated annealing and tabu search methods. The method previously presented by the author is also tested for comparison. These methods are evaluated through a computer simulation wherein problems are randomly generated. Key-Words: virtualization; optimization; metaheuristic; algorithm; cloud computing; tabu search