Performance Enhancement of MapReduce Framework in Big Data Application Using Load Balancing with Cache

Sushant Shirish Nagavkar, Ashishkumar

Abstract: Hadoop is open source software that is used to store big data, it supports data demanding applications and performs analysis, using a random placement method for parallel processing to give effortlessness and load balance. To achieve maximum parallelism per group to load balance a new Data-gRouping-AWare (DRAW) data placement is used. Problem in big data is when any query executes repeatedly it repeats whole process of execution to obtain result. In MapReduce framework and generates a large amount of intermediate data. Such huge amount of information is thrown away after the tasks finish, because MapReduce is not able to use this data. Dache, a data-aware cache framework for big-data applications gives the produced intermediate results to the cache manager. Task inquiries the cache manager before performing the actual computing work.

Keywords: BEA, Big-data, caching, DACH, DRAW, Hadoop, HDFS, MapReduce