Closure and Convergence: A Foundation of Fault-Tolerant Computing
IEEE Transactions on Software Engineering - Special issue on software reliability
Providing guaranteed services without per flow management
Proceedings of the conference on Applications, technologies, architectures, and protocols for computer communication
Proceedings of the conference on Applications, Technologies, Architectures, and Protocols for Computer Communication
Self-stabilizing systems in spite of distributed control
Communications of the ACM
A General Theory of Constrained Max-Min Rate Allocation for Multicast Networks
ICON '00 Proceedings of the 8th IEEE International Conference on Networks
Minimum Rate Guarantee without Per-Flow Information
ICNP '99 Proceedings of the Seventh Annual International Conference on Network Protocols
Scalable quality of service across multiple domains
Computer Communications
B-neck: a distributed and quiescent max-min fair algorithm
Proceedings of the 30th annual ACM SIGACT-SIGOPS symposium on Principles of distributed computing
| Hi-index | 0.00 |
Let a flow be a sequence of packets sent from a source computer to a destination computer. Routers at the core of the Internet do not maintain any information about the flows that traverse them. This has allowed for great speeds at the routers, at the expense of providing only best-effort service. In this paper, we consider the problem of fairly allocating bandwidth to each flow. We assume some flows request a constant amount of bandwidth from the network. The bandwidth that remains is distributed fairly among the rest of the flows. The fairness sought after is max-min fairness, which assigns to each flow the largest possible bandwidth that avoids affecting other flows. The distinguishing factor to other approaches is that routers only maintain a constant amount of state, which is consistent with trends in the Internet (such as the proposed Differentiated Services Internet architecture). In addition, due to the need for high fault-tolerance in the Internet, we ensure our protocol is self-stabilizing, that is, it tolerates a wide variety of transient faults.