Admission control and dynamic adaptation for a proportional-delay diffserv-enabled web server

ABSTRACT

We consider a web server that can provide differentiated services to clients with different QoS requirements. The web server can provide N > 1 classes of service. Rather than using a strict priority policy , which may lead to request starvation, the web server provides a proportional-delay differentiated service (PDDS) to heterogeneous clients. An operator for the web server can specify "fixed" performance spacings between classes, namely, ri,i+1 > 1, for i = 1,…,N - 1. Requests in class i + 1 are guaranteed to have an average waiting time which is 1/ri,i+1 of the average waiting time of class i requests. With PDDS, we can provide consistent performance spacings over a wide range of system loadings. In addition, each client can specify a maximum average waiting time requirement to be guaranteed by the web server. We propose two efficient admission control algorithms so that a web server can provide the QoS guarantees and, at the same time, classify each client to its "lowest" admissible class, resulting in lowest usage cost for the client. We also consider how to perform end-point dynamic adaptation such that clients can submit requests at lower class and further reduce their usage cost, without violating their QoS requirements. We propose two dynamic adaptation algorithms: one is server-based and the other is client-based. The client-based adaptation is based on a non-cooperative game technique. We report diverse experimental results to illustrate the effectiveness of these algorithms.

REFERENCES

• 1: V. Cardellini, E. Casalicchico, M. Colajanni, and M. Mambelli. Web switch support for differentiated services. In Performance and Architecture of Web Servers (PAWS), Boston, June 2001.
• 2: M. Crovella, R. Frangioso, and M. Harchol-Balter. Connection scheduling in web servers. In Proceedings of USITS'99, Boulder, October 1999.
• 3: C. Dovrolis, Dynamic Class Selection: From Relative Differentiation to Absolute QoS, Proceedings of the Ninth International Conference on Network Protocols, p.120, November 11-14, 2001
• 4: Constantinos Dovrolis , Dimitrios Stiliadis , Parameswaran Ramanathan, Proportional differentiated services: delay differentiation and packet scheduling, Proceedings of the conference on Applications, technologies, architectures, and protocols for computer communication, p.109-120, August 30-September 03, 1999, Cambridge, Massachusetts, United States
• 5: R. Gibbons. Game Theory for Applied Economists. Princeton University Press, 1992.
• 6: V. Kanodia and E. Knightly. Multi-class latency-bounded web services. In IEEE/IFIP IWQoS 2000, Pittsburgh, PA, June 2000.
• 7: L. Kleinrock. Queueing Systems: Vol 2. Wiley-interscience, New York, 1976.
• 8: S. C. M. Lee, J. C. S. Lui, and D. K. Y. Yau. Admission control and dynamic adaptation for a proportional delay diffserv-enable web server. In Technical Report, CUHK, Department of Computer Science & Engineering, 2001.
• 9: M. K. H. Leung , J. C. Lui , D. K. Y. Yau, Characterization and performance evaluation for proportional delay differentiated services, Proceedings of the 2000 International Conference on Network Protocols, p.295, November 14-17, 2000
• 10: Matthew K. H. Leung , John C. S. Lui , David K. Y. Yau, Adaptive proportional delay differentiated services: characterization and performance evaluation, IEEE/ACM Transactions on Networking (TON), v.9 n.6, p.801-817, December 2001 [doi>10.1109/90.974533]
• 11: S. Shenker. Fundamental design issues for the future internet. IEEE Journal of Selected areas in Communication, 13:1141-1149, 1995.
• 12: H. Zhu, H. Tang, and T. Yang. Demand-driven service differentiation in cluster-based network servers. In Proc. IEEE Infocom 2001, Anchorage, Alaska, April 2001.