Preserving quality of service guarantees with guaranteed rate fair aggregators (GRFA)

In this paper, we present a new service model for aggregate scheduling. This model combines a guaranteed rate (GR) scheduler with a guaranteed rate fair aggregator (GRFA), With fair aggregation, our model can provide guaranteed services to individual flows in the presence of flow aggregation. In addition, our mode] can utilize fully the configured bandwidth. This is not possible for traditional service models such as GR or PSRG working with FIFO aggregation [Yuming Jiang (2002), J.C.R. Benett et al., (2002)]. Subsequently, we derive an end-to-end delay bound for a network using our service model and compare the results with other models. The comparison shows that our model achieves the best performance, providing the lowest end-to-end delay bound as well as achieving high network utilization.