Efficient Topology Aware System of Network Levitated Merge for Hadoop Acceleration
Madhura Sadashiv Dhekane, Megha Borole · 2015
There are various ways to handle big data. The systems like Hadoop do this with a good efficiency with the help of MapReduce technology. Some problems with such systems are efficiency, speed, etc. To enhance the working of MapReduce technology for better results a system is presented. It is observed that in different phases of MapReduce, there are some repeating steps which can be minimized and the execution time can be reduced. Also the stages which are to be followed by each input so as to get the outputs are pipelined to achieve more efficiency and throughput. Using hierarchical topology suggested here will simplify the pipelining. Cache memory can be used to reduce number of memory cycles which will in result helps in improvement of the system.