Dependability in Peer-to-Peer Systems

A iming to ease the problems associated with making a variety of items available to multiple users over the Internet, peer-to-peer (P2P) systems create fully decentralized systems. By involving lots of peers, computational and storage capabilities increase; moreover, data replication, caching, and efficient query routing enhance overall performance. P2P systems can reduce complexity because each server deals with a smaller set of data and clients. However, P2P systems' long-running and dynamic nature creates dependability issues. In this article, we consider dependability with reference to scalability, fault-tolerance, security, and anonymity properties. Scalability and fault-tolerance depend on the number of peers and their interconnection, but security and anonymity depend on the systems' open nature and the Inter-net's communication infrastructure. The fact that any peer can contact any other in a network introduces data-integrity and privacy issues. In the not-too-distant future, self-adaptable and dependable systems will emerge from the development of nomadic technologies, mobile devices, and ubiquitous computing. Recent research has used P2P systems to organize ubiquitous environments 1 and to perform worldwide computations. 2 Because most current systems were designed to address specific application domains, similar concepts have been addressed with different terminologies and models. An output of the comparison provided in this article is an attempt to move toward common terms and definitions. Because the models underlying current P2P systems must be understood to support a thorough investigation of dependability properties, we briefly examine the most popular P2P systems and then compare how these systems address dependability. Most developers initially used P2P technology to implement file-sharing systems that spanned Internet-like environments, with a focus on searching and routing. More recently, they have used it to implement hash-map-like functionali-ties (the ability to map a key to a location) in distributed settings. P2P technology is evolving in response to peer unreliability, which continues to further the need for fault-tolerant systems. It has also become clear that effective systems require security and anonymity, so encryption and authentication techniques for large-scale decentralized systems are starting to arrive. P2P systems have further influenced research into distributed file systems, which are starting to resemble the P2P organization models. Several organizational models with different degrees of dependability have been proposed over the past few years. Before discussing specific dependability issues, though, let's look at the most representative P2P systems. They reveal the basic mechanisms that support the system properties discussed in the rest of the article. To unify …