Researches On Deep Packet Inspection In Cloud Computing

Cloud Computing is a hot topic both in industry and academia currently. Using resource integration and multilevel visualization technology, it efficiently provides users with large-scale computing resource as a service, which also has high availability, easy scalability and significant cost savings. Though Cloud Computing platforms are commonly developed and studied by IT enterprises, technology of Cloud Computing is still not mature enough to be widely applied in practice.The security issue of Cloud Computing is a main concern of most enterprise customers. Apart from traditional security strategy like encryption, it is also necessary to actively monitor and detect data streams in the platform. As one of the most efficient strategies for active detection and protection, most DPI researches are mainly concerned about optimizations in the performance over space, while Cloud Computing demands more for concurrent speed and stability. What is more, they did not consider distributed computing and user configurability in Cloud. Therefore, more work should be done in order to use DPI in Cloud Computing.Based on broad literature review, we examine both the algorithm of DPI and implementation of cloud-cooperation framework. So, considering the special requirements and unique characteristics of Cloud Computing, we study on how to apply DPI on Cloud Computing platform. Main contributions of this paper are as follows:1. Current development and characteristics of Cloud Computing are analyzed in this paper, the security issues of Cloud Computing and its'research progress are also discussed. Based on summary of current DPI mechanisms, the importance and difficulties of applying DPI mechanism in Cloud Computing are pointed out.2. We propose a new deep packet inspection mechanism: APFA, which is developed based on detailed analysis over causes of status explosion and flaws of existing solutions. Applying Asynchronous Parallelism and Heuristically Forecast Mechanism, it can take advantage of multi-core platform, offering an effective method to avoid overlapping problem and semantic assaults. Experiments under environment with artificial flows show that the average space APFA costs is 82.5% less than that of XFA and the average time is 67.1% less cost. In addition, in actual environment of Internet data stream, space cost and time cost are decreased by 64.8% and 35% respectively. 3. An intelligent cloud-cooperation framework of DPI is designed, given characteristics of Cloud Computing like distributed computing and user configurability. The overall efficiency is improved since it coordinates the configuration intelligently according to diversity of resources in the back end of cloud computing system and provides a scheduler for cooperative processing and defenses of repeated assaults. By comparing our design with traditional framework in virtual Cloud Computing Platform, we prove the advantages of cloud-cooperation framework on space and time performance.