DefinePK

Abstract

In today’s epoch of computer science storing and computing data is a very important phase. In recent days even a petabyte and exabytes of data is not adequate for storing large number of databases which contains large data sets. Therefore organizations today use concept called Hadoop which is a software framework of big data in their application. Hadoop is designed to store and process large volume of data sets consistently. While using geographically distributed data centers there may be a chance of data loss due to network link failure and node failures. Hadoop provides high reliability and scalability features. Along with it also afford faults tolerance mechanism by which the system functions properly even after a node in the cluster fails. Faults tolerance is mainly achieved using data replication and Heartbeat messages. Heartbeat messages are like acknowledgement messages between two nodes and Data replication is achieved by creating copies of same data sets into more than one computing nodes. In this paper we describe about the Hadoop architecture and its components and also describe how the fault tolerant mechanism is achieved using data replication and heartbeat messages.