Enhancing Replica Synchronization in Hadoop Distributed File System
Jyoti Kumari, Tarun Biswas, Satyanarayana Vuppala · 2018
This paper presents the fault tolerance and replica synchronization among storage server (Data-node) without the interference of metadata in Hadoop. It employs chunk list data structure which holds the information of relevant replicas stored in storage server. Whenever a client makes write request to the file stored on storage server, the replicas may become inconsistent and chances of getting fault increases. To improve the fault tolerance and replica synchronization, this paper has employed a new mechanism called Rapid replica synchronization. In this work, whenever the client puts write request, the one copy of replica will be updated first and then the remaining replica immediately gets update. During read request, the client will always get the modified data.