Routing and Clustering in Schema-Based Super Peer Networks

Peer-to-Peer networks have evolved from simple approaches like Napster (using a central index) and Gnutella (using simple broadcasting of queries) to more sophisticated approaches using distributed indices like DHTs [7], [9] or super peer networks [10]. These systems which allow to describe content with metadata assume homogeneous content/metadata; metadata schemas are fixed in advance for all participating peers. This is a useful assumption when all peers share the same type of content, but restrictive in more heterogeneous settings based on distributed information providers. In this position paper we will investigate P2P networks of more sophisticated peers, which provide and use explicit (typically heterogeneous) schema descriptions of their content. We will start by giving a short overview over schema-based P2P networks. We will then discuss how this additional schema information can be used for routing and clustering in such a P2P network, relying on a super-peer topology as well as explicit identifiers and indices for schemas, schema attributes and schema types. This work is done in the context of Edutella, a schemabased P2P network based on RDF.