A Probabilistic Approach for Achieving Fair Bandwidth Allocations in CSFQ

Abstract

The fair bandwidth allocations can isolate flows and protect well-behaved flows from ill-behaved ones. CSFQ (core stateless fair queueing) achieves the approximate fairness by dropping the extra packets beyond the fair share bandwidth at the routers. A heuristic method is used to estimate the fair share in CSFQ. Furthermore, we know that SRED (stabilized RED) uses a probabilistic method based on a Zombie list to estimate the number of flows at the router. In this paper, we take the probabilistic idea from SRED and apply it in CSFQ to estimate the fair share without using the Zombie list. Simulation results show that the new probabilistic approach achieves a comparable or even better performance than the original heuristic approach