Nowadays, network-based applications are growing rapidly such as cloud computing services, image processing, Internet of Things sensors, network traffic analysis mobility and video streaming services etc. It requires a massively distributed computation which leads to a quick and big increase in the usage of the network as well as it challenges the existing network management. Hadoop MapReduce is a framework that allows a committed and a scalable number of servers for analytics process. We can scale disk I/O requirements with the number of servers during computing and such type of scaling provides high network traffic in the underlying network of the Hadoop MapReduce. This comparative study reveals that network traffic engineering is an important research area in the MapReduce. The survey underlines the cutting edge in improving the performance of MapReduce using recent techniques and its usefulness for processing large-scale data-set. Based on this study the conclusion is that it is very difficult to change the existing network topology or network configurations frequently; therefore an Application-Aware Networks in Software-Defined Networking (AAN-SDN) is the best approach for network traffic management as well as for improving Hadoop MapReduce performance efficiency.
Volume 10 | 15-Special Issue
Pages: 202-210