Improving grouping genetic algorithm for virtual machine placement in cloud data centers

Cloud computing the newly emerged service oriented paradigm, has changed IT industry significantly. Virtualization is the main technique to empower cloud computing by separating compute environments from the actual physical infrastructure and creating virtual machines (VMs). Mapping of these virtual machines to the physical servers is called virtual machine placement problem and known to be NP hard. On the other hand, the grouping genetic algorithm which generally used for this problem does not perform efficiently in many cases. In the current work, we improve this algorithm by introducing a unique and efficient method for encoding and generating new solutions. Using vector packing problem, we model the problem of virtual machine placement and try to reduce power consumption by minimizing the number of used servers and also maximizing resource usage efficiency. The algorithm is tested over varying VM placement scenarios which show encouraging results.