ISC |
|
January 22, 2013,
Volume 59, No. 18 |
Big Data, Big Collaboration:
ISC Offers Penn Researchers High-Performance Networking Services and Collaboration Tools
From scientific discovery to business intelligence, “big data” is changing the face of research today. For the scientific, biomedical, and engineering communities, “big data” has led to “big collaboration.” Researchers across the globe now collaborate to create and analyze massive, highly complex “virtual” data repositories, uncovering the valuable predictive insights, patterns, relationships and anomalies that lie within them.
At Penn, Information Systems and Computing (ISC) is providing high-performance networking services and collaboration tools to support big data research. To meet the needs of researchers and the requirements of grant-making agencies who are increasingly seeking an institutional commitment to these services, ISC offers network infrastructure that enables the rapid transfer of large-scale data sets across campus—and across the world. Big data can stretch into terabytes—a quantity so vast that just a few years ago, it would have been faster to ship data on disk drives, rather than transfer it over the network.
PennNet, the University data network, is designed to support the majority of today’s high-performance networking needs. High-speed Ethernet connections link campus buildings and provide large-capacity external links to the Internet and Internet2, the national high-speed research and education network. ISC also offers reliable, cost-effective network solutions that can be tailored to support specific high-performance computing and data-intensive research needs. These are available as fixed bandwidth connections to remote locations, including cloud providers and data centers, or as on-demand bandwidth circuits.
A division of ISC, MAGPI, the Mid-Atlantic GigaPop for Internet2, offers national and international high-speed network connections through Internet2. For large, long-distance scientific data flows, MAGPI also provides direct national and international connections with bandwidth that can be increased as needed through Internet2’s WaveCo service. These connections provide an economical alternative to static direct connections without compromising speed or security.
With funding from the National Science Foundation, ISC and MAGPI are also offering an on-demand network capability known as DYNES (Dynamic Network System) that utilizes the Internet2 Interoperable Network (ION). DYNES/ION can create short-term, dynamically allocated circuits across many parts of the globe at a moment’s notice. Through this service, researchers can quickly and easily create linkages to distance labs and remote facilities, eliminating the lead-time for accessing high bandwidth international links and enabling the “real-time” transmission and analysis of large data sets. Currently DYNES/ION allows transmissions of up to 10 Gigabytes per second (Gbps) with a future capability of 100 Gbps.
High-Performance Networking in Action: PENTACON and the ATLAS Project
ISC’s high-performance networking services are supporting the participation of Penn scientists in two major international “big data” initiatives: the Personalized NSAID Therapeutics Consortium (PENTACON) and the ATLAS Project at the Large Hadron Collider (LHC) at the European Organization for Nuclear Research (CERN), in Switzerland. ISC and MAGPI have worked with researchers and technical staff to develop networking solutions that meet the unique needs of both projects.
To support the work of the Institute of Translational Medicine and Therapeutics at the Perelman School of Medicine as a key partner in the Personalized NSAID Therapeutics Consortium, ISC and MAGPI are providing a direct, high-speed connection between the Penn Center for Bioinformatics (PCBI)—the next generation sequencing center for PENTACON—and Princeton University’s High-Performance Computing Data Center. To enable the processing and transmission of the vast amounts of nonclinical data that PENTACON researchers are collecting, ISC and MAGPI are providing first a shared circuit, and then a dedicated dark-fiber circuit, to create a direct connection from Penn to Princeton. Currently, this connection is a 1-Gbps circuit, with the potential to increase to 10 Gbps or even 100 Gbps.
PENTACON is an international group of 42 scientists at 22 institutions led by Dr. Garret A. FitzGerald, director of PSOM’s Institute of Translational Medicine and Therapeutics. PENTACON seeks to develop an approach to personalizing chronic drug therapy by focusing initially on the factors that determine different efficacies and side effects for a single class of drugs, nonsteroidal anti-inflammatory drugs, or NSAIDs. Using the power of contemporary “omics” technologies—genomics, proteomics, metabolomics—researchers are seeking molecular signatures that discriminate drug action in cells and model systems. PENTACON collaborators will then integrate this information with clinical trial data to understand how NSAIDs impact biological networks, looking for both benefit and risk.
ISC and MAGPI have designed a high-capacity, on-demand network connection to support Penn’s participation as an analysis site for the ATLAS project at the Large Hadron Collider at the European Organization for Nuclear Research in Switzerland. ATLAS is one of the two main particle detectors at the LHC, and played a critical role in last year’s detection of the Higgs boson—the elemental particle that imbues the universe with mass. ATLAS measures the particles produced by proton collisions in the LHC and infers the existence of rare particles, like the Higgs boson.
In addition to contributing to the design and construction of ATLAS, researchers from Penn’s department of physics & astronomy, led by Dr. Brig Williams, have helped to create a process that records only the most noteworthy of the millions of proton collisions that occur in the LHC every second. Vast quantities of data—often at the petabyte scale—are then transmitted to locations around the globe, including Penn, for analysis.
Rather than create a costly direct circuit between CERN and Penn, ISC is utilizing a connection that can transmit large amounts of data, but that operates only when necessary. Using DYNES/ION capability, Penn researchers receive large amounts of data about significant proton collisions—while using only the amount of bandwidth required for each transmission. Operations and maintenance of this service are shared financially between Penn, the University of Delaware, Princeton University and Rutgers University. The economies of scale associated with this type of network access enable the rapid transmission of data at a fraction of the cost of a direct network connection.
For more information about ISC’s high-performance networking services, please contact Mark Wehrle, director of network operations at wehrle@isc.upenn.edu or visit www.upenn.edu/computing/pennnet/custom/
ISC’s MAGPI Virtual Symposium on Advanced Networks and High-Performance Computing—February 7
On February 7, 2013, from 1-4 p.m., ISC’s MAGPI will host a virtual symposium on Advanced Networks and High-Performance Computing: The Big Data Challenge. The symposium will address issues surrounding big data and scientific collaboration and present best practices for incorporating information technology in research and pedagogy. Keynote and breakout sessions are designed to demonstrate the current range of high-performance computing projects and resources, and how advancements in the high-performance computing space can be applied to teaching and research activities.
For information about the symposium, or to register, please visit: www.magpi.net/Community/Programs/Advanced-Networks-and-High-Performance-Computing-Big-Data-Challenge
The symposium will also be live-streamed for view-only access in a web browser. An archived video of the symposium will also be available after the event at video.magpi.net/
To learn more about the networking, high-performance computing and collaboration services that ISC’s MAGPI offers to researchers at Penn and in the Tri-state area, please visit www.magpi.net, or contact Greg Palmer at gpalmer@magpi.net
|