CUP provides a general purpose framework for maintaining caches of metadata in peer-to-peer networks, where continuous updates are expected, yet nodes must have personal economic incentive to participate in the maintenance. CUP is a complete protocol with query channels for coalescing bursts of queries and update channels for asynchronous delivery of query responses and updates of cached metadata. To moderate propagation without imposing a global policy, CUP introduces the notion of investment return for motivating each node to participate in the update propagation and policies for estimating when the benefit ceases to outweigh the overhead. For the case of locating content in a peer-to-peer network, we find that CUP secures an investment return of 2 to 300 times the propagation cost and significantly reduces query latency.
We have leveraged the CUP protocol to deliver metadata required for effective load-balancing of content downloads across multiple replica nodes [16]. As with regular searches, the economic incentive-based model helps to moderate and control the amount of metadata update propagation in a highly dynamic environment where load information changes very rapidly. Future work includes the use of CUP to enhance management of dynamic content replication, publish-subscribe applications, and price negotiation and auctioning of services amongst nodes in a peer-to-peer network.