While the performance characteristics of access networks and end-user-to-server paths are well-studied, measuring the performance of the Internet's core remains, largely, an uncharted territory. With more content being moved closer to the end-user, server-to-server paths have increased in length and have a significant role in dictating the quality of services offered by content and service providers. In this paper, we present a large-scale study of the effects of routing changes and congestion on the end-to-end latencies of server-to-server paths in the core of the Internet.

Although traffic between Web servers and Web browsers is readily apparent to many knowledgeable end users, fewer are aware of the extent of server-to-server Web traffic carried over the public Internet. We refer to the former class of traffic as front-office Internet Web traffic and the latter as back-office Internet Web traffic (or just front-office and back-office traffic, for short). Back-office traffic, which may or may not be triggered by end-user activity, is essential for today's Web as it supports a number of popular but complex Web services including large-scale content delivery, social networking, indexing, searching, advertising, and proxy services. This paper takes a first look at back-office traffic, measuring it from various vantage points, including from within ISPs, IXPs, and CDNs. We describe techniques for identifying back-office traffic based on the roles that this traffic plays in the Web ecosystem. Our measurements show that back-office traffic accounts for a significant fraction not only of core Internet traffic, but also of Web transactions in the terms of requests and responses. Finally, we discuss the implications and opportunities that the presence of back-office traffic presents for the evolution of the Internet ecosystem.

The knowledge engineering of belief networks is discussed. Several design issues that arose during the construction of two belief network-based systems, Pathfinder and ARCO1, are described. The issues of accuracy, consistency, and calibration as they emerged during the design of these systems are addressed, and the ways in which compatibility of all networks designed for the same domain suggests an architecture for combining the recommendations of independently designed knowledge bases into a single, consensus recommendation are discussed.>

Consensus diagnoses arise when several experts contribute their opinions about the relative merits of a series of competing hypothesis, and a single decision maker combines their responses and makes a decision without further discussion among the contributors. Consensus diagnoses were simulated by allowing an oracle to generate 'opinions' based on universal background knowledge and all available information about the specific problem being diagnosed. Contributors' opinions were generated by perturbing the oracle's opinion; the size of the perturbation depended on the contributor's degree of expertise. Several different aggregation functions were then used to reclaim the oracle's opinion from those of the contributors. The performance of these functions was compared as panel size and hypothesis-set size varied from two to ten. Comparative and individual analyses indicated that for panels assembled under circumstances similar to those of this study, small, simple methods work best.>