FAULT TOLERANT DISTRIBUTED FILE SYSTEM

Abstract

Despite distributed file systems having already been implemented for individuals and companies, users continue to face problems in using these systems and it is a still a big challenge for the researchers. The functionalities of a well functioned and an ideal distributed file system would be to provide high scalability, provide shared access to the same set of files to all its users, provide high performance, provide more storage space and more user friendliness. Google File System answers to these problems in a very efficient way and been a source of inspiration in development of this project. We are trying to simulate working of GFS for small scale data set. In this project we have implemented a Distributed File System with multiple levels of fault tolerance. At first level we are trying to filter faulty nodes/clients before they enter into system or we can say register with the server. At second level, to avoid failure we are maintaining chunk replicas so that if a client possessing file chunks gets disconnected, the file can still get downloaded. For security purpose, DES algorithm has been used for encryption and decryption. So we are successfully distributing and retrieving a file from a DFS in a very secure and fault tolerant manner. Thus, helping in load balancing by distributing the file over the network and increasing the overall system performance. Many different concepts such as transparency, fault tolerance, scalability, reliability have been defined in this dissertation and solved them in FTDFS approach. Various implementations and design strategies have been demonstrated by a survey along with different distributed systems for file sharing. Testing and performance evaluation of the system have also been carried out in this dissertation and also possible future work or the improvements have also been discussed.