High Energy and Nuclear Physics SIG Meeting
Fall 2003 Internet2 Member Meeting

The Internet2 High Energy and Nuclear Physics Special Interest Group (SIG) will meet on Tuesday, October 14 from 6:30-8:00 pm EST at the Fall 2003 Internet2 Member Meeting. The meeting will feature an update on the activities of the HENP SIG. Results of recent proposals will be discussed along with their
current status. There will be speakers on End-to-end measurement and monitoring in HENP (MonaLisa, IEPM-BW), on IPV6 for HENP, as well as characterization of end-to-end applications. Part of the meeting will review the status of our goals and plan for our next meeting. The format will be presentations followed by questions and discussion.

Here is the agenda for the meeting:

6:30 - 6:45 Welcome, Introduction and Proposal Status — Shawn McKee, University of Michigan [ppt] [html]
Welcome, agenda review and a brief report on the status of some recent
proposals relevant to our group

6:45 - 7:00 MonALISA: A Distributed Service Architecture — Iosif Legrad, Caltech/CERN [ppt] [html]
Iosif will demonstrate MonALISA, including the new 3D interface

7:00 - 7:15 Measurement and Fault-Finding using MAGGIE and PIPES — Paola Grosso, SLAC [ppt] [html]

7:15 - 7:30 IPv6 deployment at SLAC — Paola Grosso, SLAC [ppt] [html]
SLAC (Stanford Linear Accelerator Center) has recently started to implement a test IPv6 network. We will describe the characteristics of the implementation of this network and the first results of our IPv6 network monitoring.

7:30 - 7:50 Statistical Characterization of End-to-End Performance of Networked Applications — Abhijit Bose, University of Michigan [ppt] [html]
A fundamental problem in providing end-to-end performance guarantees to applications in distributed environments, such as the Internet and Computational Grids, is to characterize the application traffic in terms of both network-level and application-level parameters. An application may be composed of multiple streams, and can be anycast. Further, the underlying service platforms are increasingly built on a combination of fixed line, satelite and wireless networks. Such heterogenity introduces additional complexity of network parameters such as available bandwidth, delay, jitter and packet loss. In this talk, we present a general framework for statistical characterization of networked applications across multiple protocol layers that can identify the most important parameters affecting the end-to-end performance of the application and the underlying network, as well as provide mapping functions of their interactions. Access to such models will enable applications to respond to changing network conditions, especially in mobile and wireless environments where such adaptive behavior is crucial to maintaining user-perceived QoS levels. Such characterization methods can also help network providers design, tune and provision their networks to accommodate different classes of applications more efficiently. We demonstrate the viablity of the statistical characterization approach on an experiment testbed and show the validity of the overall approach.

7:50 - 8:00 Discussion on Goals, Status and next meeting