Performance modeling of skip models for VM migration using Xen

Cloud computing is the Internet based computing to deliver service. The major components to establish cloud are distributed systems, service oriented computing, web2.0, virtualization and utility computing. The integration of these components are required to make available data anytime and anywhere. Migration of virtual machines is the key issue to manage heterogenous cloud for load balancing. The migration is performed to accommodate the VM of highly loaded server to lightly loaded server. Pre-copy and post-copy are two migration approaches. Pre-copy is the iterative based algorithm which has better performance over post-copy due to its effective execution process for transferring pages. The outcome of pre-copy algorithm is dependent on the dirty and skip models. Dirty model is useful for identifying those pages which are being dirtied for the number of iterations. Skip model is the special case of dirty page mechanism through which consecutive iterations can avoid sending dirty pages for current iteration using skip bitmap of Xen. The skip bitmap is evaluated based on Writable Working Set (WWS) of any migration process. The proposed skip algorithm is developed to handle WWS optimally to improve performance of live migration system. The results show the state of the art to gain 91% accuracy using proposed skip model.