摘要
基于流表的转发机制为OpenFlow提供了灵活的可编程能力,但是随着网络功能的不断膨胀,OpenFlow交换机中的流表规模呈现出不断增长的趋势,这些流表难以在交换机有限的三态内容寻址存储器(TCAM)中进行存储,成为网络发展的一个瓶颈。为了高效地利用有限的TCAM资源进行流表存储,该文提出一种OpenFlow多级流表结构及其映射算法,将单一流表映射到多级流表中进行高效存储和查找。仿真结果表明:该文所提方法比单一流表的存储方法节省17%~95%的TCAM资源。这对OpenFlow数据平面查找结构及其扩展性设计具有重要意义。
Flow-table based forwarding enables flexibility on the OpenFlow data plane. However, expanding the network functions results in explosive growth of the flow table size, so that it can not be stored in the limited TCAM in the switches, so the storage becomes a network bottleneck. This paper describes an OpenFlow multiple-table pipeline architecture that efficiently stores the flow table in the TCAM memories with an algorithm that maps a flow table to multiple tables for storage and look up. Simulations show that the algorithm reduces the TCAM usage for single table storage by 17 %--95%, which is important for scalable designs of OpenFlow data planes.
出处
《清华大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2014年第4期437-442,共6页
Journal of Tsinghua University(Science and Technology)
基金
国家"九七三"重点基础研究项目(2013CB329105)
国家"八六三"高技术项目(2013AA010601
2013AA010605)
国家科技重大专项项目(2013ZX03002001)
中兴通讯产学研合作基金