Brookhaven National Laboratory

LHC T1/T2/T3 Networking Workshop

Date & Time Meeting Location
May 14-15, 2008
Brookhaven National Laboratory
Berkner Hall, Bldg 488, Berkner Room B

Today is: Saturday, April 20, 2019
Registration Is Now Closed...

Video Stream Online Video Stream of Workshop

Online video streaming service provided by the BNL Media & Communications Office and Web Services. All streams are provided in Real Media format.

Scope & Purpose
The LHC experiments at CERN will rapidly reach tens of petabytes of stored data under intense analysis. The design, construction, and data analysis require the combined intellect and dedicated work of international collaborations. However, technological limitations on storage, transmission and analysis of data impose difficult choices. For example, the LHC experiments expect to be able to record and share over networks less than one millionth of the collisions they observe. This draconic real-time selection will necessarily have to be optimized for expected new discoveries rather than the totally unexpected ones that are the dream of every physicist.

Even after the draconian selection, LHC collaborations will face the challenge of empowering thousands of geographically distributed physicists to derive physics insight from tens of petabytes of data. Although the raw cost of bandwidth is no longer a crippling impediment, the end-to-end performance of applications often needs significant improvement. Substantial effort went into middleware research to support data-intensive, worldwide collaborative science. A minimum requirement is the location-independent ability to analyze data to empower all of an experiment’s physicists to work collaboratively on datasets, growing to tens of petabytes in the next few years, using all computing resources to which they have access.

Over 68 U.S. universities and national laboratories will begin receiving data generated at the LHC later this year. Major questions facing this user community include “what impact will their science needs have on the U.S. network providers at the campus, regional, state, and national levels?”, and “what kind of performance and service level can scientists expect from service providers along the network path and the computing facilities they are transparently using?”.

The LHC Network Workshop will bring together the network engineering staffs from network providers and sites, and developers of prominent HEP production and analysis applications relevant to LHC experiments to conduct final planning and provide guidance in preparation for LHC data analysis. Topic areas of particular interest will include transfer performance optimization (protocols and end systems), end-to-end monitoring, network operation, troubleshooting and network services restoration procedures, and forecasted future network growth, among others. To address the needs that arise from bursty transfer patterns as they are foreseen in the experiment’s computing models recent research efforts into virtual circuit provisioning and QoS/MPLS capabilities will be discussed as well, given their current and envisioned future deployment into the LHC production networking environment.

Most importantly, the network management and engineering staffs at Tier-2/Tier-3 sites will have an opportunity to interact directly with the staffs of their cognizant network service providers and Tier-1 site, to address local issues of immediate concern. The intended outcome of the workshop is to provide a clear path moving forward as we prepare for LHC production, and to address any unresolved network-related issues that remain.

Scott Bradley (BNL)
Phil Demar (FNAL)
Michael Ernst (BNL)
Ian Fisk (CMS)

Information Technology Department (BNL)
USATLAS Computing Facility (BNL)

Top of Page

Last Modified: May 14, 2008