Size-based Adaptive Bandwidth Allocation: Optimizing the Average QoS for Elastic Flows

With a view on improving user perceived performance on networks supporting elastic flows, e.g., multimedia/data file transfers, we identify the key properties that an online bandwidth allocation policy should have. We then propose a family of bandwidth allocation criteria which depends on the network state, namely, the residual work of on-going transfers. Analysis and simulations show that allocating bandwidth in this fashion can improve the user perceived average bit transmission delay (BTD), i.e., delay/job size, by 50%-66% over traditional approaches. A simple implementation based upon TCP Reno, exemplifies how one might approach practically realizing such gains. In addition, we show that achieving the same performance improvements by simply adding capacity is “expensive”, requiring capacity to be augmented by about 40-50%. Moreover, for networks such as the Internet, where forecasting loads and dimensioning links is challenging, and thus transient or even persistent overloads are hard to avoid, we show that differentiating bandwidth allocations based on job size, provides an effective means to achieve “graceful” performance degradation. Keywords— Bandwidth Allocation, Fairness, QoS, Elastic Flows