Abstract :
Summary form only given. Traditional routing and traffic engineering enable network providers to stir traffic according to their economic and technical constraints. These solutions are best suited to applications like Web and Email with fixed end-points, but donpsilat interact as well with the newest breed of P2P applications. By selecting selfishly the sources that provide the fastest data rates, such applications affect actively the paths from which data flow in, and thus undermine the value of standard traffic control mechanisms. This can create a state of conflict between ISPs and P2P applications, and lead to drastic measures such as throttling or blocking of P2P connections. In this talk, I will present the potential of less radical Traffic Engineering 2.0 control mechanisms that intervene on the P2P overlay construction in order to leverage unexploited P2P service capacity that sits in more favorable network locations. Using real traffic and topology data from one of the worldpsilas largest ISPs, detailed information from 25,000 swarms with more than one million P2P clients, and a combination of analytic, numeric, and emulation methods, we show that ISPs can reclaim control of their network without severely impacting on their customerpsilas QoS.
Keywords :
Internet; peer-to-peer computing; quality of service; telecommunication network routing; telecommunication traffic; Email; P2P overlay construction; QoS; World Wide Web; network routing; standard traffic control mechanism; traffic engineering; Communication system traffic control; Emulation; Network topology; Routing; Telecommunication traffic; Traffic control;
Conference_Titel :
Broadband Communications, Networks and Systems, 2008. BROADNETS 2008. 5th International Conference on