| 摘要 |
A technique is provided for routing traffic in an integrated services network which supports a plurality of different service classes including a relative low priority class (e.g. best-effort) and a relative high priority class (e.g. guaranteed sessions). The technique of the present invention improves inter-class resource sharing efficiency and achieves high network throughput of each class of service in the network. The technique discourages guaranteed (high priority) traffic from using links that are already loaded with best-effort (low priority) traffic. This is achieved by using, in the link cost for high priority traffic, a concept of "virtual residual bandwidth", derived from the link residual bandwidth by taking into account of the congestion condition of low priority traffic. The proposed mechanism is simple in the sense that no changes are needed to the path selection algorithms employed for individual service classes the only change is in the link cost. The distribution of network resources across traffic classes adapts dynamically and automatically to the mix of traffic in the traffic classes, even if the traffic distribution is uneven across the network. When the traffic load is concentrated, significant performance improvements are achieved for low priority sessions without penalizing high priority sessions.
|
