Support for resource aggregation in collaborative P2P systems

摘要

In the last years, usage alternatives of P2P systems have been proposed, in which nodes provide resources and use resources of others in a collaborative way, in order to accomplish high scale tasks. Such systems are called “Collaborative P2P Systems” and have an important role in resource aggregation: advertisement, discovery, selection, matching, and binding phases. However, any toolkit does not allow developers to incorporate resource aggregation support into such systems using a specific programming language. In addition, existent solutions do not achieve cohesion among the key phases of resource aggregation (selection, matching, and binding). The main contribution of this paper is the design and implementation of a novel support that provides cohesion among such key phases and allows developers to aggregate multi-attribute, single-attribute, dynamic, static, and heterogeneous resources into collaborative P2P systems. The proposed Java-based support follows an unstructured topology based on super-peers, which advertise, select, match, and bind resources on behalf of their peers, reducing the network traffic. To validate our proposal, we developed a collaborative P2P application that generates the Mandelbrot set by means of the collaboration of several heterogeneous resources.