期刊名称：International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
印刷版ISSN：2278-1323
出版年度：2016
卷号：5
期号：4
页码：1198-1201
出版社：Shri Pannalal Research Institute of Technolgy
摘要：The Map-Reduce model streamlines the huge scale data handling on possessions group by abusing similar map and reduces assignments. Though numerous activities have been made to increase the execution of Map-Reduce works, they ignore the network activity produced in the shuffle stage, which assumes a fundamental part in implementation upgrade. Generally, a hash capacity is utilized to segment intermediate values among reduce assignments, which, nonetheless, is not effective of the fact that network topology and its data range connected with every key are not thought seriously about. To overcome this , Decomposition based dynamic algorithm and online algorithm is proposed to manage the huge scale optimization issue for huge information application is similarly intended to change information package and accumulation in a dynamic way. The next step to be enhanced in this method is cost and time complexity. In proposed system chooses K- medoid clustering algorithm for resource allocation which is implemented in decomposition based distribution algorithm. In the proposed work, the Euclidean distance is calculated from the dataset of latitude and longitude of the location to find the distance between locations which enhance network traffic during shuffling. This method is cost and time efficient compared to the previous works.
关键词：Map Reduce; Big Data; Data partition; ; Aggregation
