In the emerging big data scenario, distributed file systems play a vital role for storage and access of large data generated in the web-based information systems. Improving performance of a distributed file system is a very important research issue in the current context. Client side caching and prefetching techniques enhance performance of the distributed file system. Efficient replacement policy is required to improve performance of the caching process. In this paper, we have proposed a novel client side caching algorithm namely hierarchical collaborative global caching algorithm and a cache replacement algorithm namely rank-based cache replacement algorithm. We have used support value computed for the file blocks, for prefetching, caching and replacement purposes. We have proved through simulation experiments that the proposed algorithm performs better than the existing algorithms discussed in the literature.
Read full abstract