Competent Load Rebalancing For Distributed File Systems In Cloud

Main Article Content

P. Sankari
M.Bala Ganesh


Distributed file systems are fundamental factors for cloud computing applications using MapReduce Technique [1]. MapReduce algorithms are used to do searching, sorting and other operations in efficient and parallel way [2]. In distributed file systems, nodes are used for storage and computing functions. Usually file is divided into n number of chunks and chunks will be allocated to n number of nodes in cloud environment. So that MapReduce can perform between the nodes in parallel manner. However, in cloud computing environment failure may occur at any time and nodes may upgrade, add or replaced in the system. Files can also be dynamically deleted, appended and created [1]. This leads to load inequity among the nodes in a distributed system. That is file chunks may not equally distribute across the nodes. Existing distributed file system in clouds implemented based on central load balancer for chunk reallocations [1]. This dependency is completely insufficient in large scale dynamic and data intensive clouds, failure prone environment because of the overload of the central load balancer. CLRDFC against centralized load rebalance technique [1] and strongly recommends distributed load rebalancing algorithm, which taking care of load rebalancing among the nodes. So that load rebalancing task can share across multiple nodes and can avoid total system failure at a time [1]. Replication results indicate that CLRDFC is as good as with the existing central node approach and significantly doing fine than prior distributed algorithm in terms of movement cost, load imbalance factor and algorithmic operational cost.


Keywords- Distributed file systems, Load balancing, Cloud environment , chunk , node


Download data is not yet available.

Article Details