A NOVEL APPROACH TO IMPROVE PERFORMANCE OF HADOOP USING EFFICIENT DATA AWARE CACHING FOR BIG DATA APPLICATION

Continuous generation and explosion of large amount of data has focused the attention of researchers and scholars from industrial, academic area to capture the potential business opportunities and academic benefits from Big Data. The Hadoop MapReduce framework is dev eloped and widely accepted by open - source communities for solving massive parallel processing operations with the help of distributed environment. MapReduce systems in various situations are applied to achieve definite performance goals, upgrade existing systems to meet increasing business demands using novel network topology, new scheduling algorithms and resource arrangement schemes. There are various applications which need recursive operations on input data. As most of data is unchanged in an iterati ve programs reloading and reprocessing it results in wasting Input/Output, network bandwidth, and CPU resources and also put extra load for scheduling tasks, reading data from disk, and moving across the network. To avoid the load of these extra operation s, new efficient data aware caching framework is introduced with the help of cache programming model and value degree cache replacement algorithm. Data structure is created for caching to store data for temporary purpose needed by software or hardware. Res ults obtained demonstrate that efficient data aware caching decreases significant completion time of MapReduce jobs