A Model and a Design Approach to Building QoS Adaptive Systems
This chapter addresses the task of building Internet-based service provisioning systems where the quality of services (QoS) provided should not be perturbed due to changes in execution environments and user requirements. Specifically, it presents a system architecture and identifies a model appropriate for developing distributed programs that would implement the system. The model abstracts the network performance and dependability guarantees typically offered by the Internet service providers and is termed the probabilistic asynchronous model. The protocols for this model are shown to be derivable from those developed for the well-known classical models, namely: the synchronous and the asynchronous models. A protocol for reliable broadcast is derived from a synchronous protocol, together with QoS management algorithms. The system architecture prescribes the role of QoS management algorithms to be: feasibility evaluation on QoS requests from the end users, and adapting system protocols in response to changes in the environments.
KeywordsCorrect Process Failure Detector Delay Distribution Synchronous Model Reliable Broadcast
Unable to display preview. Download preview PDF.
- 1.Arvind, K.: Probabilistic Clock Synchronisation in Distributed Systems. IEEE Transactions in Parallel and Distributed Systems 5(5), 475–487 (1994)Google Scholar
- 8.Cherkasova, L., Fu, Y., Tang, W., Vahdat, A.: Measuring and Characterizing End-to-End Internet Service Performance. ACM Transactions on Internet Technology 3(4) (November 2003)Google Scholar
- 11.Di Ferdinando, A., Ezhilchelvan, P.D., Mitrani, I.: Performance Evaluation of a QoSAdaptive Reliable Multicast Protocol. Technical Report CS-TR-833, School of Computing Science, University of Newcastle (April 2004)Google Scholar
- 12.Ezhilchelvan, P.D., Mostefaoui, A., Raynal, M.: Randomized Multivalued Consensus. In: The proceedings of the fourth International IEEE Symposium on Object oriented Real-time Computing (ISORC), May 2001, pp. 195–201 (2001)Google Scholar
- 15.Gibbens, R., et al.: Fixed Point Models for the end-to-end performance analysis of IP Networks. In: Proceedings of the thirteenth ITC Specialist Seminar: IP Traffic Measurement Modelling and Management, Montrey, USA (September 2000)Google Scholar
- 16.Guerraoui, R.: Revisiting the relationship between Non-blocking Atomic Commitment and Consensus. In: Proceedings of the Ninth International Workshop on Distributed Algorithms, September 1995, Springer, Heidelberg (1995)Google Scholar
- 18.Hadzilacos, V., Toueg, S.: Fault-Tolerant Broadcasts and Related Problems. In: Mullender, S. (ed.) Distributed Systems, pp. 97–146. Addison-Wesley, Reading (1993)Google Scholar
- 21.Jacobson, V.: Congestion Avoidance and Control. In: The proceedings of the SIGCOMM symposium, August 1988, pp. 314–332 (1988)Google Scholar
- 23.Miley, M.: Reinventing Business: Application Service Providers. ORACLE Magazine, pp. 48-52 (December 2000)Google Scholar
- 27.Pias, M., Wilbur, S.: EdgeMeter: Distributed Network metering. In: Proceedings of the IEEE Openarch 2001 conference, Anchorage, Alaska (April 2001)Google Scholar