------------------------------------------------------------------------------- ******************************************************************************* ------------------------------------------------------------------------------- B O S T O N U N I V E R S I T Y Computer Science Department C O L L O Q U I U M Wednesday, April 24, 1996 3:00 pm (Coffee served at 2:30 pm) Seminar Room / MCS 135 ------------------------------------------------------------------------------- ******************************************************************************* ------------------------------------------------------------------------------- Network and Host Support for Transmission of Continuous Media David J. Yates Department of Computer Science University of Massachusetts at Amherst We are experiencing explosive growth in the global information infrastructure driven by the desire to use it for commerce, education, and entertainment. Thus, the information available through sources like the World-Wide Web is becoming richer as text and image data is being augmented with continuous media, such as voice and video. This talk focuses on two complementary issues in providing quality of service guarantees to connections sending voice and video over a global network. The first issue I examine is the end-to-end delay distribution seen by individual sessions under simple first-come first-served (FCFS) multiplexing in a network model with two significant features: (1) all traffic is connection-oriented, (2) cross traffic along routes is representative of that seen by connections in a wide area network. I compare these delay distributions with the worst case analytic delay bounds predicted by three different techniques for providing such bounds. I use our delay distribution results to examine the tradeoff between the QOS requested by a connection, the manner in which the QOS guarantee is provided, and the number of connections along a route through the network. The second issue I examine is how to provide throughput guarantees to connections on a large-scale multiprocessor server. My results show that different implementation choices are appropriate for sending continuous media voice and video when compared with conventional data, such as text and still images. For continuous media, delivering the desired rate to each connection is accomplished by a novel technique wherein threads directly measure their own throughput and feed this information to the scheduler. ------------------------------------------------------------------------------- For colloquium info, including directions, see http://cs-www.bu.edu/colloquium For more information contact Prof. Mark Crovella -------------------------------------------------------------------------------