Adaptive Replication control in HDFS supported supervised learning
K. Sree Divya, Abhinav Kumar, T Vijaya Rao · 2017
Apache Hadoop could be very famous now-a-days.The applications based totally on ApacheHadoopare growing in nowadays due to its lively and dierent functions.Hadoop dispensed file gadget(HDFS) is the heart of Apache Hadoop which is reliable and exceptionally to be had.it'sfar having static Replication approach by way of default.In utility Layer,there are parallel operations had been achieved on the le because of which get admission to price for each records record becomes dierent.This static Replication approach ends in deadly errors at the performance of the gadget.toovercome the disadvantage of HDFS,This paper explains the dynamic approach to duplicate information files based on Supervised studying.theuse of probability Distribution,each records le is used to create corresponding Replication approach.throughmeasuring access capacity,excessive potential files were going for Replication and final Low potential les are entered in thesection of ErasureCode.because of this dynamic technique,this paper indicates specific Replication approach and Erasure Code Mechanism to improve Availability and Reliability.