Resource discovery is one of the most challenging problems in P2P networks, in particular for structured DHT-based P2P networks. Different techniques have been studied, thoroughly analysed, compared and contrasted as part of this project, and as a first and mandatory step prior to the development of our approach to the problem.

There is a close relationship and future opportunities for collaboration between the Grid and P2P communities. Resource discovery is a traditional key component in Grid architectures whose design can be extremely improved thanks to the application of P2P technologies. On the other hand, structured and unstructured P2P technologies are becoming more and more popular on new application fields. P2P is moving away from classic file sharing systems towards general global-scale applications where the ability to efficiently manage resources much like in Grid environments is mandatory.

Our new approach to the resource discovery problem is motivated by the special requirements identified when designing a decentralized and distributed QoS-aware caching architecture to be deployed in a set of clusters built from cheap off-the-shelf computers. The design of the resource discovery service is layered over a single underlying DHT overlay network which connects all the peers in the cluster and is based on the combination of (1) a spanning-tree built mapping DHT peers to their parents in a bottom-up fashion, and (2) a set of routing indices which allow peers to efficiently lookup other peers in the DHT overlay network, matching some resource constraints.