Load balancing between nodes in a volunteer cloud computing by taking into consideration the number of cloud services replicas

摘要

In this paper, we investigate the load balancing between nodes in the volunteer cloud computing. We propose a new approach which is based on the idea of cloning a cloud service on one or more nodes when the number of the user requests will be important at a given time. Our solution allows a better system reliability and reduces the response time of the users by distributing their requests between the volunteer nodes. Unfortunately, the replication of cloud services minimizes the storage space capacity. Thus, we propose a second algorithm that selects and deletes the replicas of a cloud service without degradation of the load balancing, using for this the Markov Chain Models. The experimental results, based on PeerSim simulator, show that the proposed algorithms can effectively achieve good performance (load balancing) and improve the response time.