Issue link: http://osercommunicationsgroup.uberflip.com/i/265921
S u p e r C o m p u te r S h o w D a i l y 1 7 Tu e s d a y, N o ve m b e r 1 9 , 2 0 1 3 SCSD: Are you introducing any new products? SC: At SC13 we are announcing two new HPC products: Our 3U High-Density Computation Accelerator (HDCA) holds up to 16 boards within the external chassis, pro- viding up to 74TB per system. The sys- tem ships with either NVIDIA Tesla GPUs (such as the new K40) or Intel Xeon Phi cards pre-installed and tested. Our Flash Storage Array (FSA) is also a 3U rackmount unit, and can hold up to 32 flash memory cards. When configured with the latest Fusion-io cards, the FSA provides 200TB of high-speed storage. Performance for this FSA has been meas- ured at up to 8 million IOPs. SCSD: Describe your company's current marketing strategy. SC: As a relative newcomer into the HPC market, OSS has chosen to team with HPC industry leaders to bring these new products to market. This has the added advantage of full server and accelerator testing and support. Our partners include IBM, Exxact, Penguin and SGI. We are also working closely with NVIDIA, Intel and Fusion-io to ensure full product com- patibility with their add-in cards. SCSD: What distinguishes your products from the competition? SC: The primary alternative to the OSS One Stop Systems ( Cont'd. from p. 1) external expansion approach is to add the compute accelerator or flash add-in cards into a host server. For one or two card implementations, this approach can be cost-effective. However, as additional add-in cards are needed, this internal approach often breaks down. Servers tend to be constrained by the number of slots, power and cooling capacity. Pushing servers to the limits can result in less reliable solutions that operate at reduced performance. Moving the add-in cards to an external chassis provides sub- stantially more slots, power and cooling. The external chassis approach uti- lizes a high-performance host card that installs in the host server. In this way, physically smaller and lower cost servers can be utilized, and still have access to large quantities of compute and flash storage resources. SCSD: What is the outlook in general for this product line? SC: We believe that the HDCA and FSA are the future within HPC. The OSS strat- egy leverages our strengths in PCIe over cable to help users utilize more of these key technologies to dramatically increase the performance of their HPC systems. By teaming with key industry leaders such as IBM, Intel and NVIDIA, OSS is expanding the limits of HPC. During SC13, visit One Stop Systems at booth 1137. For more information, visit www.onestopsystems.com, call 760-745-9883 or email scooper@onestopsystems.com. rather than competitive in nature. We work closely with many research organ- izations and universities around the world to develop our open-source tools. For our commercial customers, this means we can offer cost-effective pro- prietary solutions on top of our open- source platforms. SCSD: What technologies are you focus- ing on at SC13? BG: We are highlighting our HPC and advanced visualization efforts. Specific capabilities that we're demonstrating include an immersive VR demo with ParaView, where attendees can interact with a 3D dataset; visualization and analysis for nuclear engineering; and computational chemistry, among others. We're showcasing our range of technical expertise and the types of commercial challenges we can solve. SCSD: Are you doing anything beyond your exhibit booth? BG: Kitware's booth is a stop on Intel's "Fellow Traveler Tour," and we are teaching a ParaView tutorial as part of the SC13 technical program that covers the architecture of ParaView and the fun- damentals of parallel visualization. At the tutorial, attendees will learn the basics of using ParaView for scientific visualiza- tion with hands-on lessons. The tutorial also features detailed guidance in visual- izing the massive simulations run on Kitware ( Cont'd. from p. 1) today's supercomputers and an introduc- tion to scripting and extending ParaView. SCSD: Are you introducing any new products at SC13? BG: We're not doing a product launch at the conference, but we are more broadly introducing one of our new initiatives, VTK-m. SCSD: What is VTK-m? BG: VTK-m is the next-generation analysis and visualization framework that addresses the challenge of exascale data, which we are developing in col- laboration with the DOE National Labs. VTK-m will efficiently harness the parallel capabilities of future processors, and appropriately accom- modate the architectures needed for future parallel computing. SCSD: If our readers are interested in learning more about these efforts, how should they get in touch? BG: Kitware will be at booth 4207 all week, so stopping by is probably the fastest way to get information on our services or opportunities for collabora- tion. Our website, www.kitware.com also has information on our scientific and other focus areas. Visit Kitware at booth 4207 during SC13. For more information, visit www .kitware.com, call 518-371-3971 or email kitware@kitware.com. SA: Ciena's booth will feature packet networking and converged packet optical portfolios. Attendees can see our SDN testbed in action, with a software con- trolled network responding in real-time from the StarLight switching facility in Chicago. Attendees can also tour our mobile demonstration lab and visit with Ciena experts to discuss strategic archi- tecture initiatives such as Ciena's path to terabit transmission, high-scale Ethernet2 switching solutions and 100G submarine networking. Ciena is also powering SCinet, the network linking SC13 to the outside world. We are supplying a 400G high- capacity coherent optical networking link, combined with Layer 2 terabit-scale packet switching. SCSD: Elaborate on Ciena's SDN testbed. SA: Along with Internet2, ESnet, CANARIE and StarLight, we are build- ing the industry's first fully operational end-to-end WAN that leverages OpenFlow across the packet, OTN circuit and photonic transport layers. It is sup- ported by an open architecture carrier- scale controller and intrinsic multi-layer operation, and includes real-time analyt- ics software applications. Researchers can use the testbed to trial new technologies without having to build a unique infrastructure for every use case. SCSD: How have Ciena's customers needs changed during the last decade? SA: Networks are on the verge of funda- mental change. Bandwidth demand is exploding, driven by new types of servic- Ciena ( Cont'd. from p. 1) es, content and applications. This is driv- ing a seismic shift in end-user behavior and network usage. In the past, networks were about connecting locations. With the rise of cloud computing and the Internet of Things, we are at an even greater inflection point, where the next wave is about connecting devices and machines. That kind of connectivity is less static and unpredictable, and our cus- tomers face the challenge of architecting their networks to accommodate that. In the area of high-performance computing and networking, we are seeing the real- ization of open, dynamic, large-scale net- works that are leveraging software-based control technologies, driving conver- gence and flattening of architecture. It's a "new game" for the network. We help our customers to not just add bandwidth, but architect their networks to be more programmable so they can grow their business. SCSD: Please explain Ciena's OP n architecture. SA: We introduced the term OP n to highlight the principles on which our platforms and technologies are devel- oped. OP n combines optical and pack- et networking technologies with soft- ware and open interfaces to help providers build networks with greater scale, programmability and network- level applications. OP n enables R&E institutions to push the boundaries of research through unmatched network capacity. It also enables support for dynamic application demands with a path toward SDN. During SC13, visit Ciena at booth 1924. For more information, go to www.ciena.com, call 800-207-3714 or +44 20 7012 5555 or email pr@ciena.com. TERABIT NETWORKING GETS ITS CLOSE-UP AT SC13 CERN, the European Organization for Nuclear Research, reports that experiments at the facility, which include the Large Hadron Collider, have generated 75 petabytes of data during the last three years. The volume of information coming out of CERN is enormous, and a far-reach- ing network of data centers and supercom- puters store and analyze files, while teams of scientists collaborating across conti- nents work to understand and make sense of these high-end experiments. CERN is just one example of how advanced scientific research has become globalized and data-intensive. Here at SC13, the world is getting a preview of how the technology community is responding to meet the demand for increased capacity for scientific processing power, be it for particle physics, medical research or environmental monitoring. With bandwidth-hungry scientific projects needing to send more data over longer distances faster than ever, one of the big trends at SC13 this year is the evolution of terabit networking. SCinet, the high-capacity network underpinning the data flows within SC13 and also linking the convention center to the outside world, pulls in advanced tech- nology each year to build out a network designed to showcase the current state of networking innovation and meet the needs of dozens of high-performance computing demos and workshops. Year after year, SCinet increases the capacity of the show network at Supercomputing, and this year is no exception. Leveraging expertise from recent terabit networking trials, Ciena is helping SCinet assemble a 400G network that will link the conference site to the various global networks outside. Ciena and SCinet are also providing terabit- scale layer two switching within the local area conference network. Momentum is quickly building towards 400G and beyond, and SCinet is demonstrating that the technology is (nearly) ready for primetime. The team at CERN can expect the next 75 petabytes to move much quicker than the previous 75. During SC13, visit Ciena at booth 1924. For more information, go to www.ciena.com, call 800-207-3714 or +44 20 7012 5555 or email pr@ciena.com.