Analytic Evaluation of Quality of Service
for On-Demand Data Delivery
Hongfei Guo
guo@cs.wisc.edu
Haonan Tan haonan@cs.wisc.edu
Abstract:
Quality of service (QoS) measured as balking probability and average
waiting time is of great interest for on-demand data delivery service
providers. In this study, we develop and validate two analytical models, a
balking model and a waiting model, for evaluating balking probability and
average waiting time respectively for media servers with limited
bandwidth. Based on in-depth analysis of simulation results, we also
propose an interpolation of the average service time in the waiting model
when measuring the Patching protocol. Compared to simulation, the balking
model captures the trend of balking probability with changing server
bandwidth; the waiting model yields reasonably accurate results when
measuring the Hierarchical Multicast Stream Merging protocol; the
interpolation improves the accuracy of the waiting model when measuring
the Patching protocol. Finally, using those two models, we evaluate and
compare two multicast protocols - Patching and HMSM - and suggest server
bandwidth C* as a reasonable trade-off between QoS and server utilization.