Comparative Analysis of Adaptive File Replication Algorithms for Cloud Data Storage

With the popularity of cloud computing technology, storage systems becomes an important role to distribute and store massive amount of data to data centers effectively. As grow as the explosive amount of data in cloud environment day after day, data popularity in cloud computing becomes unstable and unpredictable. Moreover, maintaining the static number of replicas in the system results highly storage cost for unpopular data and inefficient for most accessed data. To address these challenges, adaptive file replication algorithm called PopStore approach is proposed by using data access history and popularity thresholds. The performance of the replication algorithms is analyzed by using Yahoo Hadoop audit log to extract data access pattern. According to the experimental results, the PopStore strategy and LALW are able to adapt the access frequency and data popularity. In addition, PopStore can save more storage cost than LALW and network usage can be saved up to 35% more than LALW strategy.