The potential of the P2P concept for designing the next-generation of real-world distributed applications can be realized only if a comprehensive framework quantifying the performance related aspects of all classes of P2P applications is available. However, only content-sharing applications based on the P2P concept have reached the desired level of maturity.
P2P networks have caught the imagination of the research community and application developers with their sheer scalability and fault-tolerance characteristics. The proposed framework would hopefully lead to quantifiable Service-Level Agreements for a variety of peer-to-peer services and applications. The individual performance measures which comprise the QoS framework are also discussed in detail along with some thoughts on how these can be complied with. Early results from the prototype implementation of the Peer Enterprises framework (a cross-organizational P2P collaborative application) are used as a basis for formulation of the QoS parameters. Hence, this research paper proposes an early QoS framework covering various classes of P2P applications content distribution, distributed computing and communication and collaboration. Researchers have proposed some QoS (Quality-of-Service) parameters for content-sharing P2P applications based on response time and delay, but these do not cover the gamut of application domains that the P2P concept is applicable to. The potential and social cost of equilibrium without strategy restrictions are also investigated.Įventually, we propose an algorithm which fully exploits the advantages of strategy restrictions. In particular, this is done in terms of strategy restriction The minimum and congestion is also minimized. Hence we look for those equilibria where streaming length is at Still, these profiles are not sustainable as equilibria,īecause they also result in null congestion, while selfish optimization by individual players has to result The whole population in a minimum number of steps. Of directed trees specifying how each content unit reaches each peer) that distributes the whole content over Large set of fastest streaming strategy profiles, each providing a streaming tree (i.e. Who to ask for some content unit at each date (over a finite, discrete time-horizon). With strategy spaces containing time-sequences of agents as elements, because each player has to choose We model this as a (intrinsically dynamic) congestion game Has to rely upon P2P (peer-to-peer) exchanges. Sends each content unit to some agent as soon as its production has been completed, and next distribution In real-time streaming some content is to be distributed across agents while being produced.
0 kommentar(er)
