Scalable Fair Random Early Detection
Xiaohui S. Lin, Kaiyu Zhou, Hui Wang, Gong-Cao Su · 2006
This paper proposes the scalable fair random early detection (SFRED) to improve the fairness of RED. As a router generally sees packets from a fast flow more often than a slow flow, this suggests that the fairness of a RED router can be improved without per-flow information. In SFRED, fair bandwidth allocation is attained with a list that stores statistics of limited active flows. Based on this list, SFRED takes identifying and punishing the fast, unresponsive fast, and protecting slow flows into consideration. Simulations with simple IP networks demonstrate that the fairness of RED has been significantly improved with the implementation of SFRED. Different from the previous proposals the complexity of SFRED is proportional to the size of the list but not coupled with the queue buffer size or the number of active flows, so it is scalable and suitable for various routers