Author Email: firstname.lastname@example.org
The exponential growth in the usage of cloud environment with increased usability of data and the explosion of data from various sources lead to the data replication in cloud computing. Cloud is a popular platform for replication of recurrently used data which is emerged as a prevalent alternative to the traditional cluster based replication to sustain growing demand of data and to provide high data availability & improved performance. This research article will consider the parameters such as number of inbound user’s requests, diversified structures of cloud storage, accessible bandwidth and wavering degree of popularity of the contents to investigate and provide comparative analysis of available replication strategies. The data in demand is replicated on the desired node in the cloud environment which leads to better data availability, faster data access and optimized utilization of bandwidth. The replica is created if the content popularity crosses the threshold, to retain the high availability of the data in the system. When the popularity of the data reduces with the dropping demands, the stale replica needs to be withdrawn from the data node. This paper proposes a method of creation and withdrawal of the replica depending on the content popularity threshold to provide an effective replication strategy and also to provide increased availability with optimal storage consumption in the data nodes.
Keywords Replication, Data Availability, Cloud Computing, Content Popularity, Replicas.
Two most popular and widely practiced approaches in computer world namely distributed systems and database management system are combined with the data replication. Data replication is an approach that allows maintaining multiple of copies of particular data known as “Replicas” at several data nodes to provide better access in the distributed system. Performance improvement, fault tolerance system and wide availability of data can be achieved by replicating the datasets or files. The data which is recurrently accessed by users is replicated on several sites to provide easy and prompt access to all the users located at different sites without any constraints. The data to be replicated will be preferably read only data. This also enables the availability of the data in demand on multiple sites at the same instance. The requested data can be made available to the user at the closer site leading to improved performance and reduced communication overhead.
This paper has proposed a data replication strategy which can be utilize the storage space in the cloud environment efficiently. The methodology proposes a withdrawal of replica module depending on the popularity of the block. It also gives a comparative analysis of existing data replication strategies based on the results for reduction in storage space & cost, increased availability & reliability, optimized bandwidth utilization, minimized number of replicas & response time etc. The analysis also gives a perspective to the advantages and drawbacks of the available strategies. These drawbacks are explored to propose a strategy to SEDReS (Storage Effective Data Replication Strategy in cloud environment) with a provision of withdrawal of replica of replica once it is no more in demand. As a future scope the proposed methodology will be implemented using a simulator tool and will be compared with the existing strategies for performance.
378 total views, 2 views today