Published Paper Details:

Dynamic Data Allocation and Redistribution model for Heterogeneous Hadoop Clusters for different level of constraints to represent its interaction on the data in adoption to the virtualisation. The importance of the proposed methodology is to minimize the completion length (i.e., makespan) of a set of MapReduce jobs. The current Hadoop only permits static slot configuration, i.e., fixed numbers of map slots and reduce slots all through the lifetime of a cluster. Dynamic configuration may lead to high system resource utilisations as well as less completion length. It is considered as effective schemes which use slot ratio between map and reduce tasks as a tunable knob for reducing the makespan of a given set leveraging the workload information of recently completed jobs, schemes dynamically allocates resources (or slots) to map and reduce tasks. The corresponding constraints have been placed to that particular node for effective data allocation to the specified streaming data. Further it eliminates the data duplication occurring in the data clusters. In addition incremental clustering eliminates the inconsistent instance considered as an outlier would be processed effectively.