Enhancing NameNode fault tolerance in Hadoop over cloud environment

Risks on user applications has significantly changed in the cloud computing infrastructure because of the failures such as server overload, network congestion, hardware faults, etc. Hadoop uses a single metadata server namely NameNode (NN) which keeps the whole file system metadata in primary memory. In Hadoop, the I/O operations associated with those credential metadata makes NameNode a Single Point of Failure. The system achieves fault tolerance by using an Observer Tool, which will continuously monitor the NN. It proactively calculates chances of any crash and in such a case system will run using the secondary NN.