A SURVEY ON MAPREDUCE IN CLOUD COMPUTING

Abstract

Author(s): Dr.M.Newlin Rajkumar, S.Balachandar, Dr.V.Venkatesakumar, T.Mahadevan

Cloud Computing delivers computing resources so that it was appealing great thoughtfulness. Map Reduce is a programming model that be linked with the implementation for treating and creating large data sets. Since there is many drawbacks such as low scalability, does not support edible pricing and stream data processing etc. So to overcome the all drawback of Map Reduce framework, Cloud Map Reduces (CMR) is proposed. The experimental results of CMR show that it is more efficient and development faster than other implementations of the MR method. Also CMR can be improved to Support stream data processing, edible pricing using Amazon Cloud’s spot instances. Improve speed-up to process over traditional MR that processes more than 30% for large data sets and provides flexibility and scalability. CMR is aimed for handling batch data with major modifications are made in the basic structure of CMR. The pipelining between Map and Reduce phases for supporting flow data processing is introduced here also previous feature of CMR is now called as Continuous Cloud MapReduce (C-CMR). The architecture of CMR consists of various components such as Simple Storage Service, Input/Map queue, Simple database, Map Workers, Combiners; etc. CMR is suitable to start on a Map Reduce job since the nodes are symmetric.