报告时间：2018年11月20日周二15:30

报告地点：湖南大学信息科学与工程学院106教室

报告人简介：沈潼，2013年至在湖南大学信息科学与工程学院从事博士研究，导师张大方教授，2016年-2018年，在中科院计算所客座访学，导师为谢高岗研究员。主要研究兴趣为未来网络，数据包分类，路由查找。主要成果发表在JCST，IEEE ISPA, IEEE Globecom, IEEE IPCCC等国际期刊和会议上。

报告主要内容：Packet classification according to multi-field ruleset to guide packet forwarding is a key component for network appliances. SDNs and cloud networks update the rulesets frequently to reflect policy changes. Supporting fast ruleset updating and high-speed packet classification is thus mandatory for these networks. Nevertheless, existing packet classification studies focus either on high-speed packet classification (e.g., decision tree-based algorithms) or on fast updating (e.g., hash based algorithms), and no known approaches meet both requirements.

In this paper, we propose Range-tuple Space Search (RTSS) to effectively accelerate the packet classification in hash based algorithms while keeping their strength in fast rule updating. RTSS is built on our key observation that rules are unevenly distributed in the tuple space, which is defined by the combinations of prefix lengths of fields to match.

To reduce the number of hash tables for fast classification, we introduce a novel concept range-tuple with each formed as the aggregate of adjacent tuples. RTSS can overcome the major obstacle that hinders hash-based packet classification by balancing the number of hash tables and the probability of hash collision.