Abstract: Cloud basically deals with huge volumes of data and each cloud user stores and accesses gigabytes of data. Cloud applications that require very fast data access are emerging quite frequently. Hence, it is of utmost importance to provide a high performance with reliability and efficiency. The Hadoop Distributed File System is one of the most widely used distributed file systems. Such cloud systems must provide efficient caching mechanisms in order to improve the speed of accessing the data. However, due to inefficient access mechanisms of the Hadoop, the access latency is too high that it reduces the throughput of the system. An efficient pre-fetching technique in Hadoop improves the overall performance of the cloud system.