Free Trial

Safari Books Online is a digital library providing on-demand subscription access to thousands of learning resources.

Share this Page URL

Chapter 32: Trust and Fairness Managemen... > FAIRNESS MANAGEMENT IN P2P AND GRID ... - Pg. 759

Trust and Fairness Management in P2P and Grid Systems games, the arbiter is trusted not only to perform the verification correctly but also not to reveal the obtained information to other players. A similar approach using trusted third parties to supervise transactions has been proposed by (Liu et al., 2004). fAIRNeSS MANAGeMeNT IN p2p AND GRID SYSTeMS The problem of fairness management in P2P and grid systems concerns the fairness of access and provision of shared resources. In P2P file sharing, these resources are the access bandwidth of peers. In grids, resources can be CPU time or, more generally, processing time of grid tasks. Unfair behavior in P2P and grid systems is often called free-riding. Peers or grid nodes can sometimes use resources of others without providing resources in return. The goal of fairness management is the increase of fairness in the distribution of used and provided resources. The two distributions are often combined using a distribution of the ratio of used and provided resources. In P2P systems, fairness management works by providing incentives for peers to provide resources to the system. The de- centralization of control in P2P systems makes it difficult to manage fairness. Peers can only attempt to control the behavior of others using individual strategies, like in a non-cooperative game. P2P fairness management considers the following measures of fairness of individual peers: sharing ratio and altruistic provision. Trust Management in web Services and Grid Systems Grid systems often use Web services to implement frameworks of grid services. The Web services architecture includes a special Trust Management method that relies entirely on recommendations to establish trust, and does not calculate reputation. The method also relies on trusted third parties: authorities that can issue recommendations to agents who wish to use a Web (or grid) service. In the Web services TM approach, an agent obtains certain recommendations (called security tokens) after authentication from an authority. When the agent wishes to invoke a Web service, she will present her recommendations. The process of obtaining and presenting recommendations is standardized by the WS-Trust standard. On the other hand, an agent that provides a service (for example, a member of the grid) can express her preferences concerning recommenda- tions. Various recommendations can be required in order to have access to various services. The requirements of a providing agent can be specified using rules described in the WS-Policy standard. When a requesting agent presents recommenda- tions, the providing agent checks whether her policy is satisfied, and based on this check makes the decision whether to provide the requested service. The TM method used in Web services and grids can be exploited to support workflows, yet this requires better methods for trust negotiation and retrieval of trust recommendations (Blaze et al., 1996). TM methods can also be used for access control in Grid systems (Colombo et al., 2007). fairness Management in p2p file Sharing Sharing ratio is defined as the total number of uploaded bytes divided by total bytes downloaded (Mol et al., 2008). A P2P Fairness Management systems' primary objective is to keep the sharing ratio above a certain level for every peer. A shar- ing ratio of 1 and above is considered fair, since it indicates that a peer uploads the same amount (or more) as she has downloaded. If the sharing ratio drops below a certain level it is possible to exclude the peer from the system. Altruistic provision is defined as the differ- ence between the expected upload rate and the download rate (Mol et al., 2008; Piatek et al., 2007). Piatek et al. also give an alternative defini- 759