Improving cloud computing virtual machines balancing through hosts and virtual machines similarities

Quality of service is one of the major concerns in cloud computing. Virtual machines (VMs) balancing techniques can help reduce service degradation in cloud computing environments. Several works have presented cloud computing balance techniques; however, only a few used the similarity between VMs and physical hosts to map VMs migrations. In addition, most proposals do not consider the size, dynamism, and heterogeneity of the cloud when developing a management technique. We present a cloud computing VMs balancing algorithm that uses the similarity between VMs and physical hosts to create the map of migrations. Furthermore, the proposal takes into account the size, dynamism, and heterogeneity of the cloud when mapping VMs migrations; thus the proposal is developed in a distributed fashion, enabling the processing of each cluster at a time. To evaluate the proposal, we used the Google cluster data set. Experiments demonstrate that the proposed technique can improve the balance of allocated resources; thus helping reduce service degradation. Moreover, the runtime of the algorithm indicates that it is feasible to be used in a real cloud computing environment with hundreds of physical servers and virtual machines.