N. Mansouri, M. M. Javidi,
Volume 15, Issue 3 (9-2019)
As grow as the data-intensive applications in cloud computing day after day, data popularity in this environment becomes critical and important. Hence to improve data availability and efficient accesses to popular data, replication algorithms are now widely used in distributed systems. However, most of them only replicate the static number of replicas on some requested chosen sites and it is obviously not enough for more reasonable performance. In addition, the failure of request is one of the most common issue within the data centers. To compensate these problems, we, propose a new data replication strategy to provide cost-effective availability, minimize the response time of applications and make load balancing for cloud storage. The proposed replication strategy has three different steps which are the identification of data file to replicate, placing new replicas, and replacing replicas. In the first step, it finds the most requested files for replication. In the second step, it selects the best site by consideration of the frequency of requests for replica, the last time the replica was requested, failure probability, centrality factor and storage usage) for storing new replica to reduce access time. In the third step, the replacement decision is made in order to provide better resource usage. The proposed strategy can ascertain the importance of valuable replicas based on the number of accesses in future, the availability of the file, the last time the replica was requested, and size of replica. Our proposed algorithm evaluated by CloudSim simulator and results confirmed the better performance of hybrid replication strategy in terms of mean response time, effective network usages, replication frequency, degree of imbalance, and number of communications.As grow as the data-intensive applications in cloud computing day after day, data popularity in this environment becomes critical and important. Hence to improve data availability and efficient accesses to popular data, replication algorithms are now widely used in distributed systems. However, most of them only replicate the static number of replicas on some requested chosen sites and it is obviously not enough for more reasonable performance. In addition, the failure of request is one of the most common issue within the data centers. To compensate these problems, we, propose a new data replication strategy to provide cost-effective availability, minimize the response time of applications and make load balancing for cloud storage. The proposed replication strategy has three different steps which are the identification of data file to replicate, placing new replicas, and replacing replicas. In the first step, it finds the most requested files for replication. In the second step, it selects the best site by consideration of the frequency of requests for replica, the last time the replica was requested, failure probability, centrality factor and storage usage) for storing new replica to reduce access time. In the third step, the replacement decision is made in order to provide better resource usage. The proposed strategy can ascertain the importance of valuable replicas based on the number of accesses in future, the availability of the file, the last time the replica was requested, and size of replica. Our proposed algorithm evaluated by CloudSim simulator and results confirmed the better performance of hybrid replication strategy in terms of mean response time, effective network usages, replication frequency, degree of imbalance, and number of communications.
