Class-of-Service identification is a very useful task for enterprise, traffic, being a significant portion of the network traffic today, constitutes a highly desirable class for identification. Accurate classification of P2P traffic is a challenging problem, and becomes even more challenging when we are constrained to use only transport-layer header information. In this paper, we present a new approach for P2P traffic identification, that uses fundamental characteristics of P2P protocols, such as a large network diameter and the presence of many hosts acting both as servers and clients. We do not use any application-specific information, but we are, however able to identify both known and unknown P2P protocols in a simple and efficient way.

joined Google in 2003 and is currently a principal software engineer in the Infrastructure Security Group. His areas of interest include computer and network security, as well as large-scale distributed systems. He serves on the Usenix board of directors. MOHEEB ABU RAJAB

IOA is a formal language for describing Input/Output automata that serves both as a formal specification language and as a programming language (Garland et al. in http://theory.lcs.mit.edu/tds/ioa/manual.ps, 2004). The IOA compiler automatically translates IOA specifications into Java code that runs on a set of workstations communicating via the Message Passing Interface. This paper describes the process of compiling IOA specifications and our experiences running several distributed algorithms, ranging from simple ones such as the Le Lann, Chang and Roberts (LCR) leader election in a ring algorithm to that of Gallager, Humblet and Spira (GHS) for minimum-weight spanning tree formation in an arbitrary graph ( rithms is significant for two reasons: (a) it is an indication of the capabilities of the IOA compiler and of its advanced state of development, and (b) to the best of our knowledge, these are the first complex, distributed algorithms implemented in an automated way that have been formally and rigorously proved correct. Thus, this work shows that it is possible to formally specify, prove correct, and implement complex distributed algorithms using a common formal methodology.

No abstract

Web-based malware attacks are more insidious than ever. What can be done to stem the tide?