High-Speed DPI Method Using Multi-Stage Packet Flow Analyses

Deep packet inspection (DPI) is becoming more im- portant as an effective way of providing application-aware ser-vices and improving bandwidth efficiency. In DPI, URL analysis is key because HTTP is used for most network applications. In large-scale networks, high-speed URL matching with a large-scale URL list must be capable of handling various applications and content. It must also support wildcard expressions in the URL list to achieve flexibility and compactness. We propose a high-speed matching method that uses the hierarchical structure of the URL to support wildcard-matching through multi-stage analysis. Early hash-matching stages reduce the number of executions of the following full-text-search stage by extending the levels in the URL hierarchy. Our evaluation results show that our method is 500 times faster than the conventional binary search method in certain situations.