Self-Configured Fair Queueing

摘要

Many core-stateless mechanisms have been proposed to resolve the scalability problem in the core routers. However, they place considerable reliance on the edge routers or anterior core routers to mark correct flow information in corresponding packets. As a result, the condition of incorrect flow information may severely degrade the fairness among different flows. To achieve the robust fairness, an approximately core-stateless fair sharing mechanism, self-configured fair queuing (SCFQ) is proposed in this paper. This mechanism is mainly composed of two rate estimators; one estimator is used to estimate the arrival rate of the malicious flows that can be adopted to emend the incorrect flow information of the arrival packets. The other one is used to estimate the fair share rate. Computer Simulation is applied to compare the fairness of the SCFQ, core-stateless fair queuing (CSFQ), linear fair sharing estimation (LFSE), deficit round robin (DRR), and first in first out (FIFO). From the simulation results, the SCFQ is capable of supporting much better fairness than that of other compared mechanisms under a variety of traffic conditions. Also, the overhead of the SCFQ is low, so it is very suitable to be deployed in high-speed core routers.