Next-Generation Peer-to-Peer Systems

The Internet has changed from a statically, television-like webpage space to a highly heterogeneous and extremely dynamically information platform. No other technology forced that development in the last few years like Peer-to-Peer. Those systems often contain Millions of nodes, terabytes of data and cause most of total Internet traffic nowadays. Despite their enormous performance power they are often reduced to the filesharing context. This limitation leads to a Peer-to-Peer discussion which focuses on terms like copyright and ownership and overlooks the power such networks accommodateon the technical site.

This lecture tries to overcome that limitation and will discuss concepts for Peer-to-Peer Systems that go beyond filesharing networks. Today all Peer-to-Peer applications can be clutched into three areas: Filesharing, Distributed Computing and Communication.

When we’re looking for new Peer-to-Peer application fields we definitely miss advanced services that go beyond communication and sharing of data or computer power. In contrast to Napster, the first prominent example of Peer-to-Peer, today’s systems do not employ any centrally stored index systems. Although it’s quite obvious Peer-to-Peer offers a lot of advantages compared to traditional client-server systems such as performance and cost reduction, Peer-to-Peer developers are confronted with quite a lot new challenges that traditional Web-Developers never had to handle.

The lecture will discuss such challenges for developers like manageability, extensibility, fault tolerance, security and scalability. We will look at concepts like DHT’s for advanced searching, Replication Schemes for improved Data Management and Reputation Systems for Ring of Trusts. We will look at new emerging metadata concepts like Edutella and the adaptation of current standards like XML into the Peer-to-Peer context. While security became a highly discussed topic in distributed networks, Peerto-Peer Systems do not only need advanced security mechanisms, Peer-to-Peer technology can also help developing distributed services like Anti-Spamfilters.

A special case will be a semantic approach to the problem of information management in such huge, highly dynamic and heterogeneous databases where we will discuss a traditional Data Mining concept, Association Rule Mining, in the context of Peer-to-Peer Systems to build novel kind of services. Agentorientated systems are strongly related to that semantic approach, where autonomous agents can act as peers, form hybrid sub networks and offer services.

Finally the lecture will discuss the connection between Peer-to-Peer and Mobile Solutions. We will identify proposing mobile applications that have to integrate Peer-to-Peer concepts or Peer-to-Peer-like architectures and discuss challenges for developers.