An Adaptive Partitioning Technique to Improve the Performance of Bigdata

The performance of parallel computing purely depends on the data partitions. Hadoop is the framework to perform the computations on the partitioned data across many systems and produce the results. It also suffers with the statistic partitioning concept present in the framework over large datasets. In this paper we propose a dynamic partitioning algorithm which improves the data analytics over large datasets. Our algorithm provides user friendly reports on the given dataset and reduces the cost of the project. This new algorithm improvises the effective utilisation of the nodes in the cluster and reduces the execution time.