Title :
A High-Performance IO Engine for SDN Controllers
Author :
Sae Hyong Park ; Byungjoon Lee ; Jisoo Shin ; Sunhee Yang
Author_Institution :
ETRI (Electron. & Telecommun. Res. Inst.), Daejeon, South Korea
Abstract :
In software defined networking (SDN), a controller acts as the "brain," providing a centralized view of the whole network. Through the controller, the network can be quickly and easily programmed and managed. Consequently, the importance of an SDN controller\´s performance is significant. In SDN controllers, high performance has two aspects: fast response time, and high flow-setup rate. In this paper, we present a high-performance IO engine that gives at least a 26% higher flow-setup rate than current state-of-the-art controllers while maintaining the fastest response time. The IRIS IO engine adopts a multi-queue design, which is fundamentally different from the other controllers\´ approaches. Further, our evaluation results show that the IRIS IO engine has the best processing-core scalability among the fastest open-source controllers.
Keywords :
computer network management; public domain software; software defined networking; IRIS IO engine; SDN controller performance; flow-setup rate; high-performance IO engine; multiqueue design; network management; network programming; open-source controllers; processing-core scalability; response time; software defined networking; Engines; Iris; Libraries; Message systems; Scalability; Servers; Throughput; Controller; OpenFlow; Performance; SDN;
Conference_Titel :
Software Defined Networks (EWSDN), 2014 Third European Workshop on
Conference_Location :
Budapest
DOI :
10.1109/EWSDN.2014.30