Leveraging traffic repetitions for high-speed deep packet inspection

摘要: Deep Packet Inspection (DPI) plays a major role in contemporary networks. Specifically, datacenters of content providers, the scanned data may be highly repetitive. Most DPI engines are based on identifying signatures packet payload. This pattern matching process is expensive both memory and CPU resources, thus, often becomes bottleneck entire application. In this paper we show how can accelerated by leveraging repetitions inspected traffic. Our new mechanism makes use these to allow repeated skipped rather than again. The consists slow path, which frequently strings identified stored dictionary, along with some succinct information for accelerating process, where traffic byte but from if encountered, skipped. Upon skipping, path recovers state it would have been had scanning continued byte. solution achieves significant performance boost, especially when same source (e.g., website). experiments that such cases, our throughput gain 1.25–2.5 times original throughput, implemented software.