Abstract: A distributed file system often forms a central building block for cloud computing based on Map-Reduce Programming paradigm. The nodes in file system simultaneously serve computers in network and perform storage operations. During processing the file gets partitioned into multiple number of chunks so that Map Reduce task can be performed in parallel over the nodes. Often nodes fail when they upgrade, replace or add. File operation such as create, delete and append results in load imbalance in DFS. Uniform distribution of file chunks gets bit difficult in distributed file system. The centralized load balancing technique includes central load balancer which emerges with bottleneck problem. Proposed system focuses on load rebalancing problem in distributed file system and aims to design a load rebalancing algorithm with linear technique approach to allocate file chunks appropriately to system considering low cost, less execution time. The system will also include Hadoop Distributed File System for maintaining logs and data sets.
Keywords: Cloud Computing, DFS (Distributed file system) Load Balancing, Linear Programming, Hadoop Distributed File System.