A Physics-Style Approach to Scalability of Distributed systems
Is it possible to treat large scale distributed systems as physical systems? The importance of that question stems from the fact that the behavior of many P2P systems is very complex to analyze analytically, and simulation of scales of interest can be prohibitive. In Physics, however, one is accustomed to reasoning about large systems. The limit of very large systems may actually simplify the analysis. As a first example, we here analyze the effect of the density of populated nodes in an identifier space in a P2P system. We show that while the average path length is approximately given by a function of the number of populated nodes, there is a systematic effect which depends on the density. In other words, the dependence is both on the number of address nodes and the number of populated nodes, but only through their ratio. Interestingly, this effect is negative for finite densities, showing that an amount of randomness somewhat shortens average path length.
Unable to display preview. Download preview PDF.
- 2.Mitchell, D., Selman, B., Levesque, H.: Hard and easy distributions of sat problems. In: AAAI 1992. Proceedings Tenth National Conference on Artificial Intelligence, vol. 873, pp. 459–65 (1992)Google Scholar
- 7.Stoica, I., Morris, R., Karger, D., Kaashoek, M., Balakrishnan, H.: Chord: A Scalable Peer-to-Peer Lookup Service for Internet Applications. In: ACM SIGCOMM 2001, San Deigo, CA, pp. 149–160 (2001)Google Scholar
- 8.Stoica, I., Morris, R., Karger, D., Kaashoek, M., Balakrishnan, H.: Chord: A Scalable Peer-to-Peer Lookup Service For Internet Applications. Technical Report TR-819, MIT (2002), http://www.pdos.lcs.mit.edu/chord/papers/chord-tn.ps
- 9.Mozart Consortium: The mozart homepage (2003), http://www.mozart-oz.org
- 10.Alima, L.O., El-Ansary, S., Brand, P., Haridi, S.: DKS(N; k; f): A Family of Low Communication, Scalable and Fault-Tolerant Infrastructures for P2P Applications. In: The 3rd International Workshop On Global and Peer-To-Peer Computing on Large Scale Distributed Systems-CCGRID2003, Tokyo, Japan (2003), http://www.ccgrid.org/ccgrid2003