Approach to geographic based namespace load distribution using symphony

Abstract

With the enormous growth of volume of data, Namespace distribution and replication are becoming a major challenge as Namenode server needs to coordinate with data nodes for its status, job execution and data blocks operations. As the data is growing day by day, Hadoop Usage of Namespace impose big challenge to achieve high scalability, availability and single point of failure. The proposed approach of hadoop is based on Symphony and has resolved the issues of namespace scalability, availability and address the issue of single point of failure. The focus of work is based on geographic based namespace distribution using Symphony. By distributing the metadata storage, the Hadoop distributed file system has gain immense storage capacity by eliminating the bottleneck of the single NameNode. The approach provides the ability to dynamically add Namenodes as required, thereby increasing the underlying performance and scalability. The multiple NameNode approach contributes to distribute the workload of the Hadoop file system and eliminate the single point of failure. A comparison of Chord and Symphony has been provided and it has been analyzed that Symphony works better for geographic based key distribution technique on Namespace server.