Dedinski, I. and Meer, H. and Han, L. and Mathy, L. and Pezaros, D. P. and Sventek, J. S. and Zhan, X. Y. (2009) Cross-layer peer-to-peer traffic identification and optimization based on active networking. In: Active and programmable networks. Lecture Notes in Computer Science . Springer Verlag, Berlin, pp. 13-27. ISBN 9783642009716Full text not available from this repository.
P2P applications appear to emerge as ultimate killer applications due to their ability to construct highly dynamic overlay topologies with rapidly-varying and unpredictable traffic dynamics, which can constitute a serious challenge even for significantly over-provisioned IP networks. As a result, ISPs are facing new, severe network management problems that are not guaranteed to be addressed by statically deployed network engineering mechanisms. As a first step to a more complete solution to these problems, this paper proposes a P2P measurement, identification and optimisation architecture, designed to cope with the dynamicity and unpredictability of existing, well-known and future, unknown P2P systems. The purpose of this architecture is to provide to the ISPs an effective and scalable approach to control and optimise the traffic produced by P2P applications in their networks. This can be achieved through a combination of different application and network-level programmable techniques, leading to a cross-layer identification and optimisation process. These techniques can be applied using Active Networking platforms, which are able to quickly and easily deploy architectural components on demand. This flexibility of the optimisation architecture is essential to address the rapid development of new P2P protocols and the variation of known protocols.
|Item Type:||Contribution in Book/Report/Proceedings|
|Subjects:||Q Science > QA Mathematics > QA75 Electronic computers. Computer science|
|Departments:||Faculty of Science and Technology > School of Computing & Communications|
|Deposited On:||17 Aug 2012 16:38|
|Last Modified:||13 Feb 2014 00:05|
Actions (login required)