Reaearch And Analysis Of Hadoop Ddos Attack Detection

In recent years, the Internet industry has entered a period of rapid development. With the rapid development of mobile Internet, e-commerce and people’s constant pursuit of technology, cloud computing, which is a new industry, appeares. As more and more enterprises and organizations beginning research and analysis on cloud computing, the development prospects of cloud computing are getting better. As an open source cloud computing framework, Hadoop gradually becomes the most popular cloud computing platform. While Hadoop has been more and more concerned about by people and gradually becomes focus of research at home and abroad, because of the time it appeared is not long, there is rarely research on how dose Hadoop face the increasingly rampant network DDoS attacks. Thus, the research for Hadoop DDoS attack detection in this thesis should have some practical significance.On the basis of researching on Hadoop security and DDoS attack, the paper presents a distributed detection framework for Hadoop DDoS attack detection. In this framework, all nodes in Hadoop form a security detection union in a distributed manner and each node in the union should carries out DDoS attack detection. Secondly, the paper studies the detection algorithm used in the detection node of the league. The paper theoretically analyzes Naive Bayes, K-NN, decision trees, artificial neural networks and support vector machine those fifth algorithms.Combining with the actual test data we find that SVM is the most suitable algorithm for distributed Hadoop DDoS attack detection. In addition, for the SVM algorithm, the paper presents a Hadoop based distributed multi-process grid search method to find the optimal combination of parameters of SVM algorithm. The actual test data shows that compared with the traditional grid search, this method can complete searching the optimal combination of parameters of SVM algorithm in a short time without reducing accuracy rate. Finally, we have an implementation and experience of the distributed Hadoop DDoS attack detection framework. Experimental data show that the system can effectively detect DDoS attacks Hadoop suffering.