Junyi Deng, Yanheng Liu, and Jian Wang, Lin Li
Cloud computing is becoming a powerful parallel data processing method and it can be adopted by many network service providers to build a service framework. Though the cloud computing is able to efficiently process a huge amount of data, it is easy to be attacked due to the massive distributed cluster nodes. In this paper, we propose a novel Secured MapReduce Framework (SMRF), which establishes a close relation between the Speculative Execution (SE) and the security of the YARN. SMRF launches the speculative executions in a certain ratio, computes and compares their respective MD5 hashes of the intermediate and final results in the MapReduce process. Moreover, the proposed framework is able to discover the actual and potential malicious nodes in the Hadoop cluster. In addition, a prototype framework, called SecMR, is implemented based on Hadoop 2.3.0. The theoretical derivations and experiments show that the proposed SecMR not only guarantees the security of the MapReduce process, but also successfully locates two types of the malicious nodes in Hadoop while just increasing a little overhead.
Cloud computing, Hadoop, MapReduce, Speculative execution, Security
Junyi Deng, Yanheng Liu, and Jian Wang, Lin Li, A Novel MapReduce Framework for Improving Security of Cloud Computing. 2019 International Computer Science and Applications Conference (ICSAC 2019). 2019: 76-79.