文章基本信息

标题：TOWARD SECURE COMPUTATIONS IN DISTRIBUTED PROGRAMMING FRAMEWORKS: FINDING ROGUE NODES THROUGH HADOOP LOGS
本地全文：下载
作者：N. MADHUSUDHANA REDDY ; Dr. C. NAGARAJU ; Dr. A.ANANDA RAO 等
期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2017
卷号：95
期号：23
页码：6398
出版社：Journal of Theoretical and Applied
摘要：MapReduce programming paradigm is used to process big data in large number of commodity computers where parallel processing is leveraged. In this kind of programming phenomenon, users do not have control over distributed computations. Therefore it is quite natural for users having privacy and security concerns. The nodes involved in MapReduce computations may become malicious and cause security issues. Different attacks are possible on distributed environment. The nodes that cause such attacks are known as rogue nodes. In this paper a methodology is proposed based on analysis of Hadoop log files to find rogue nodes that are malicious and launch attacks for disturbing normal functioning of MapReduce framework. In other words, the methodology aims at integrity verification of computations in MapReduce environment. This is achieved without having any cryptographic primitives or other computational operations. Hadoop logs and low-level system calls are utilized and correlated in order to obtain operations performed by different nodes. This knowhow is then matched with system calls and invariants in program to find out malicious operations and the rogue nodes that cause such operations. The proposed methodology is evaluated with real Hadoop cluster environment to demonstrate proof of the concept. The empirical results revealed the significance of using this approach towards secure computations in distributed programming frameworks while processing big data.
关键词：Mapreduce; Hadoop; Detection Of Rogue Worker Nodes; Hadoop Logs; System Calls