Internet2

close
Use Internet2 SiteID

Already have an Internet2 SiteID?
Sign in here.

Internet2 SiteID

Your organization not listed? Create a local account to use Internet2 services.

Create SiteID

News

U.S. Large Hadron Collider ATLAS Community to Adopt perfSONAR Network Performance Suite

Posted on Oct 14, 2008 by Doug Howell
Tags: Big Data, Datasets for Research, High Energy and Nuclear Physics, imported, Network Performance, Performance Monitoring, Performance Tools, Research Funding Support

Critical Cyberinfrastructure Technology to Support High Energy Physics Research; NSF Grant To Accelerate U.S. Development and Adoption

New Orleans, LA – October 14, 2008 – When the Large Hadron Collider, the largest global scientific experiment, officially begins operations in 2009, multiple terabytes of data per second will flow out of CERN in Europe to thousands of researchers spread across the globe. In addition to continuous multi-gigabit data flows between LHC sites, scientists will each leverage international research networks to download or transmit, about two terabytes of data for analysis within four-hour windows, every few weeks. Because of these time limitations, maintaining optimal network performance on the entire network path is absolutely critical.

For this reason, the U.S. LHC ATLAS community led by the U.S. Department of Energy's Brookhaven National Laboratory has committed to adopting the perfSONAR-PS networking performance suite of services among its ten participating institutions. This deployment marks the first major implementation of perfSONAR in the U.S. and validates its role in providing important cyberinfrastructure to support advanced scientific research. ATLAS is one of six detectors in the LHC.

“perfSONAR provides users and network engineers a window into the network to generate near real time performance traffic monitoring and visualization,” said Brian Tierney, computer scientist, ESnet. “Its ability to provide global analysis of network performance problems across network domains makes it possible for users to instantly pinpoint choke points and make immediate adjustments to their applications to vastly enhance network performance and facilitate greater network reliability.”

perfSONAR is developed through a global collaboration consortium led by ESnet, GÉANT, Internet2, and RNP and in the U.S. via partnerships between the University of Delaware, ESnet, Fermi National Accelerator Laboratory, Internet2, and the Stanford Linear Accelerator Center (SLAC).

Understanding network performance issues has posed a number of challenges for distributed collaborators like those involved in the LHC. A typical physical path from one scientist to a remote instrument or collaborator requires crossing several administrative domains including campus, regional, and backbone networks. Each of these networks maintains control over who can access its diagnostic and performance data making the ability to isolate the faulty network segment, and to evaluate different solutions both challenging and time consuming.

perfSONAR provides the LHC community the resources necessary to address and resolve these issues. U.S. LHC ATLAS sites will deploy the new ‘Performance Node LiveCD,’ which allows organizations to seamlessly install a perfSONAR-enabled node with tools for monitoring passive network metrics and running active performance tests. Internet2 and ESnet are also deploying a number of hosts running perfSONAR services on their respective networks to help monitor LHC data transfers.

"The LHC represents the first major scientific endeavor that completely relies upon the use of advanced networks to support physics analysis carried out by thousands of globally distributed researchers who are key to the project's success. Networks enable these researchers to transparently utilize tens of thousands of processors and tens of millions of gigabytes of data storage that are provided by centers around the globe, resources that are seamlessly integrated by grid middleware such as the one developed by the Open Science Grid project in the U.S," said Michael Ernst, director of the RHIC and ATLAS computing facility at Brookhaven National Laboratory.

Ernst added, "perfSONAR provides the comprehensive set of tools our community needs to efficiently diagnose and address network issues across the entire network path instantly. In doing so, we drastically reduce the administrative burdens and time constraints of manually troubleshooting network issues and allow our scientists to focus their resources on doing their science rather than worrying about the network."

LHC is just the first of many application communities that are increasingly relying upon networks to create large and physically distributed scientific work environments. Today, large or specialized machines are required to continue making basic discoveries, and limited funds means that instead of scattering multiple machines around the globe, a single machine is built with remote access in mind. The promise is that these scientists will find it just as easy to work in this globally distributed environment as they used to in their own lab.

Recognizing the critical role network performance plays in enabling this next wave of scientific collaboration and innovation, the National Science Foundation (NSF) recently awarded the University of Delaware in collaboration with Internet2 and the Stanford Linear Accelerator Center (SLAC) a grant to support the refined development of perfSONAR to make it more easily deployable by the U.S. research community like those involved in LHC. The grant also provides resources to enable developers to continue their participation in the international perfSONAR consortium as well as funds for technical training, documentation, and community outreach.

“perfSONAR is an important collaborative activity and represents a major step toward achieving interdomain end-to-end network performance awareness - this comprehensive visibility has the potential to have a lasting impact across all networked scientific applications,” said Kevin Thompson, program officer for NSF's Office of Cyberinfrastructure. “NSF is pleased to support the contributions by the University of Delaware, Internet2 and SLAC in their collaborations with other research network organizations like ESnet and GÉANT."

Martin Swany, assistant professor of computer and information sciences at the University of Delaware and lead principle investigator for the grant, added, “Our collaborators have successfully worked together over the past four years to develop and begin initial deployments of perfSONAR. The NSF grant enables us to greatly elevate these efforts by providing the resources to broaden our scope of work, accelerate development and deployment efforts, and build strong community support for this important cyberinfrastructure technology.”

For more information, visit:
http://www.internet2.edu/performance/pS-PS/index.html
Information on perfSONAR will be presented today, October 14 at 3pm CDT at the Internet2 Fall Member Meeting Performance Update session in New Orleans, LA. For more information or to watch via netcast, visit:
http://events.internet2.edu/2008/fall-mm/sessionDetails.cfm?session=10000158&event=911

Media Contact:

Lauren Rotman
lauren@internet2.edu
202-331-5345