OSHEAN eCurrent | Page 30

matter particles, to particles that can give of our particle physicists are leading and researchers incredible insight into extra- transforming Big Data analytics globally.” dimensional theories. These massive data samples and the gathering and analytical Brown uses distributed computing as a primary technologies used to create and evaluate method to analyze data. It would be unrealistic them are providing insights heretofore to analyze all the data on site at its Providence unattainable. This goes beyond the physics campus, so the jobs have to run not only on with researchers exploring using tools that are on the cutting Brown’s own computers We are training our students but on computers that are connected through edge of the revolution not only in our primary goal known as Big Data. of physics and discovering (which facilitates access Brown University’s how our world works to the to distributed high- research is producing utmost detail, but also in throughput computing many tens of terabytes Big Data analytics. Many of of data a year. For the Open Science grid for research in the U.S.), the LHC grid, or others our particle physicists are that are interconnected. experiments are leading and transforming Big When Brown evaluating thousands Data analytics globally. researchers submit jobs on their cluster, data institution from simultaneously investing in before have the limits been pushed as they are – Meenakshi Narain Professor of Physics, Brown University can be used globally in its own massive supercomputing and petabyte today. With scientific computing and Big Data, any of its collaborative storage systems. high bandwidth and reliable connections are universities throughout “This is an exciting model of collaborative example, its of trillions of collisions and looking for a few hundred particles as a result — so the sta- critically important, as evidenced in this area of science where massive amounts of information the world through the grid protocol. computing and analysis,” said Greg Landsberg, access to the data is critical, as are good Apart from distributed computing, Brown Thomas J. Watson, Sr. Professor of Physics, statistical algorithms to analyze the data in also participates in the Any Data, Any Time, Brown University. “Collaboration is one of order to glean potentially critical small signals Anywhere infrastructure. Combining several the key reasons that Brown is part of CMS Brown University physicists are currently from the big noise. As a result, particle physics existing software products, AAA presents a experiment. While we all have individual using about 22 petabytes of data per year is a great training site for data researchers, and “data federation,” or a global and unified view contributions to make to the experiment and produced at CERN by the CMS experiment. many physics students are leading data science of storage systems. Many of those participating in analyzing the data, it is a collaborative In 2011/2012, the university upgraded to groups around the U.S. and globally. in the CMS experiment are utilizing the AAA experiment where we are all working together connect to the OSHEAN 10GB network, which infrastructure; as a result, Brown can run a job to produce the result.” has helped with the ongoing research. With the tistical analysis involved is great. Fast “We are training our students not only in our primary goal of physics and discovering how our world works to the utmost detail, but also in Big Data analytics,” said Meenakshi Narain, Professor of Physics, Brown University. “Many 30 | CURRENT SPRING 2017 are transferred between the U.S. and Switzerland daily. upgrade, in the last year alone, data rates are interactively on its clusters and connect to data at a collaborating institute. This highly THE ROLE OF A GOOD NETWORK now at 10-30 TB at 3-4 GBS per second. collaborative model in distributed computing Research has always been a major driver to the With its distributed analysis system, Brown alleviates the requirement for each research development of advanced networks, and never can quickly move the data it needs to discover Stronger Together | 31