DESIGN OF FILE SYSTEM ARCHITECTURE WITH CLUSTER FORMATION ALONG WITH MOUNT TABLE

Sheetu Sharma, Vikas Gupta

Abstract: Architecture for comprehensive dataset is defined as a File System. In a comprehensive cluster set, there are a large number of servers where data are directly stored. Cluster is used to store tuples from one or more relations physically closed to other in the database. Clustering is a way of storing data on a disc. In this proposed research work the file system architecture is maintained with cluster formation and mount table specification. The cluster formation is an intelligent formation which works on keyword based feature analysis on files. The related files are kept in one cluster. Along with this, the mount table is attached which is a table, that stores the keyword information as well as other metadata related to each file contained in the system. The proposed research work is extended in two main phases. In the first phase, the distributed architecture is defined in clusters. Once the architecture is defined, in the second phase the user query is filtered and the keyword is extracted from it. An extracted keyword of query is used in the hash table, to find corresponding cluster having all related files. Then all files related to that keyword are displayed. Then upon selection of one particular file, the whole content of that file is displayed. This research work is focused on solutions to get or retrieve data from large data source in less time and at less computation cost.

Keywords: Hadoop Distributed File System, Query Optimization, Indexing Technique, Distributed System, Cluster.

DOI: https://doi.org/10.15623/ijret.2014.0306077