Enhancing the Hadoop Performance through Data Placement in Heterogeneous Hadoop Cluster

In the present world large volumes of data are getting generated and these records and data details have to be maintained for future purpose. Keeping these large bulks of data and using them becomes difficult. To overcome this and make it easy to store, use and work with it a tool called Hadoop is used. Hadoop uses the concept of a cluster that is many small nodes together form a cluster. Nodes with varying configurations (like varying RAM sizes, processors) form a heterogeneous cluster. Data placement technique in heterogeneous cluster is complicated. The data placement technique in heterogeneous cluster helps in the efficient use of resources and when combined with the MapReduce programming model increases the performance. Data placement can be done by forming racks. In this work we enhance the performance of Hadoop in heterogeneous cluster by first creating racks for data placement and then modifying certain parameters of Hadoop tool. The techniques are implemented and evaluated in Hadoop 1.0.3.