ADVA Optical Networking and the University of Michigan are demonstrating how 100G technology can support large-scale sharing of scientific data and advanced data-intensive Research and Education (R&E) projects at SuperComputing 2012 (SC12). During the show, which takes place in Salt Lake City until November 15, ADVA Optical Networking and U-M are hosting a series of demonstrations using the ADVA FSP 3000 to connect servers at the ATLAS Great Lakes Tier-2 in Ann Arbor to Chicago over Michigan LambdaRail (MiLR) infrastructure, a high-speed network created jointly by the University of Michigan, Michigan State University, and Wayne State University. Advanced R&E community Internet2 is also participating by transporting the 100G signal from Chicago to the show floor in Salt Lake City. The demo highlights how this technology can be used to share resources and bring together multiple institutions and communities, facilitating complex big science applications and ground-breaking collaborative projects.
“MiLR provides low-cost, 10Gbit/s fiber optic connections to national and international research and education connection points, which facilitates collaboration to researchers throughout the world,” explained Roy Hockett, Network Architect of ITS Communications Systems at the University of Michigan. “Most big science cannot be done at a single institution; the work and information is distributed, and resources are shared. An example of this is U-M’s ATLAS Collaboratory Project, which aims to advance technologies and practices required for the organization and execution of modern, large-scale collaborative research experiments. As the process of discovering the Higgs Boson demonstrated, we need consolidation and interaction with multiple institutions to reap the benefits of big scale science.”
The ATLAS particle physics experiment at CERN in Geneva, Switzerland has over 2,500 members spread across the globe and links to the Large Hadron Collider (LHC) through a global grid infrastructure that connects partnering universities into national and regional centers. ATLAS collects massive amounts of data from the LHC, detecting and recording approximately a billion collisions per second, a data rate equivalent to twenty simultaneous telephone conversations by every person on the earth.
“This demonstration showcases how high-performance networks can support large, long-distance scientific data flows in leading data-intensive science programs,” said John Scherzinger senior VP sales for North America, ADVA Optical Networking. “The Large Hadron Collider, for example, produces around 15 petabytes of data every year. Sharing this massive amount of data with physicists all over the world requires significant computing power and a network powerful enough to handle it.”
Last year at SuperComputing 2011, Caltech HEP team along with University of Victoria and University of Michigan used the ADVA FSP 3000 within MiLR’s infrastructure to demonstrate how petabytes of data can be effectively transferred using highly tuned servers. This year’s demonstration will provide a preview of what is possible for upcoming data-intensive applications utilizing high-performance network infrastructures. It showcases a FTP 100 connection between Brookhaven National Labs (the US ATLAS Tier-1 site) and the University of Michigan’s ATLAS Great Lakes Tier-2 (a US ATLAS Tier-2 site) across a 100G infrastructure using newest generation servers. The FTP 100 project is a part of the DOE end-to-end 100Gbps network program, which aims to provide high-performance data transfer for a wide range of applications, climate simulation, bioinformatics, and data migration among cloud computing centers.