Critical Mass of a Distributed End-System Monitoring Service.
Michael Finkenzeller, Gerald Kunzmann, Andreas Kirstädter, Rüdiger Schollmeier · 2006
Abstract — The Peer-to-Peer paradigm to utilize resources at the edge of a network offers a wide area of application scenarios based on services like file sharing and GRID computing. In this paper, we employ the P2P paradigm to provide a monitoring service for network statistics to end-users, network operators or interested applications. The basic principle is that end-user devices monitor local observable data e.g. bandwidth consumption and packet loss. This data is then organized in a P2P-based distributed directory. In this paper we estimate the potential of such a service by applying it to a simple topology discovery problem. The question we answer is the number of peers and connections needed to discover a certain percentage of network topology and how long this will take. To answer that question, a simulator was implemented that is composed of different abstraction layers. It allows estimating a critical mass of peers that is necessary to cover a given percentage of random network topologies.