Heterogenous Networks Can Be Unstable at Arbitrarily Low Injection Rates
A distinguishing feature of today’s large-scale platforms for distributed computation and communication, such as the Internet, is their heterogeneity, predominantly manifested by the fact that a wide variety of communication protocols are simultaneously running over different distributed hosts. A fundamental question that naturally poses itself for such common settings of heterogeneous distributed systems concerns their ability to preserve or restore an acceptable level of performance during link failures. In this work, we address this question for the specific case of stability properties of greedy, contention-resolution protocols operating over a packet-switched communication network that suffers from link slowdowns. We focus on the Adversarial Queueing Theory framework, where an adversary controls the rates of packet injections and determines packet paths. In addition, the power of the adversary is enhanced to include the manipulation of link slowdowns. Within this framework, we show that the composition of LIS (Longest-in-System) with any of SIS (Shortest-in-System), NTS (Nearest-to-Source) and FTG (Furthest-to-Go) protocols is unstable at rates ρ > 0 when the network size and the link slowdown take large values. These results represent the current record for instability bounds on injection rate for compositions of greedy protocols over dynamic adversarial models, and also suggest that the potential for instability incurred by the composition of two greedy protocols may be worse than that of some single protocol.
KeywordsInjection Rate Heterogenous Network Link Failure Network Queue Instability Bound
Unable to display preview. Download preview PDF.
- 3.Alvarez, C., Blesa, M., Serna, M.: The Impact of Failure Management on the Stability of Communication Networks. In: Proc. of the 10th Int’l Conference on Parallel and Distributed Systems, pp. 153–160 (2004)Google Scholar
- 5.Barlow, R., Proschan, F.: Statistical Analysis of Reliability and LifeTesting Models. Holt, Rinehart and Winston, New York (1975)Google Scholar
- 8.Borodin, A., Ostrovsky, R., Rabani, Y.: Stability Preserving Transformations: Packet Routing Networks with Edge Capacities and Speeds. In: Proc. of the 12th Annual ACM-SIAM Symposium on Discrete Algorithms, pp. 601–610 (2001)Google Scholar
- 10.Diaz, J., Koukopoulos, D., Nikoletseas, S., Serna, M., Spirakis, P., Thilikos, D.: Stability and Non-Stability of the FIFO Protocol. In: Proc. of the 13th Annual ACM Symposium on Parallel Algorithms and Architectures, pp. 48–52 (2001)Google Scholar
- 12.Gary, J.: Why do computers stop and what can be done about it? In: Symposium on Reliability in Distributed Software and Database Systems (1986)Google Scholar
- 14.Koukopoulos, D.: The Impact of Dynamic Link Slowdowns on Network Stability. In: Proc. of the 8th Int’l Symposium on Parallel Architectures, Algorithms and Networks, pp. 340–345 (2005)Google Scholar
- 18.Koukopoulos, D., Mavronicolas, M., Spirakis, P.: Instability of Networks with Quasi-Static Link Capacities. In: Proc. of the 10th Int’l Colloquium on Structural Information and Communication Complexity, Carleton Scientific, pp. 179–194 (2003)Google Scholar
- 21.Tsaparas, P.: Stability in Adversarial Queueing Theory, M.Sc. Thesis, Computer Science Department, University of Toronto (1997)Google Scholar