The early peer-to-peer applications eschewed commercial arrangements and instead established a grass-roots model in which the collection of end-users provided their own distributed computational infrastructure. While this cooperative end-user approach works well in many application settings, it does not provide a sufficiently stable platform for certain peer-to-peer applications (e.g., DHTs as a building block for network services). Assuming such a stable platform isn't freely provided by a benefactor (such as NSF), we must ask whether DHTs could be deployed in a competitive commercial environment. The key issue is whether a multiplicity of DHT services can coordinate to provide a single coherent DHT service, much the way ISPs peer to provide a completely connected Internet. In this paper, we describe various approaches for DHT peering and discuss some of the related performance and incentive issues.
