CLOUD

By Steve Fisher, Editor In Chief -- In the coming weeks Supercomputing Online will bring you a series of articles on emerging companies in the high performance computing, networking and communications industries. The first article in the series focuses on Teraburst Networks, a developer of high performance optical networking solutions and visualization products. ---------- Sunnyvale, CA-based TeraBurst Networks is a developer of high performance optical networking solutions that are designed to enable users to realize the promise of next-generation photonic networking while maintaining the performance benefits of current OEO technology (optical-electrical-optical). TeraBurst's OMS systems and SubNetwork Management System are designed to provide significant cost savings and enable revenue-generating optical services. Currently, switches with electrical cores (or optical-electrical-optical -OEO) are available and offer the benefits of 3R regeneration (retransmit, reshape and retime) and performance monitoring for signal quality. However, these non-blocking switch architectures are usually based on small channel-count devices and multi-stage Clos structures and do not scale easily to large port counts. Also, they switch at lower STS-n levels with an overall fixed STS-1 capacity; this delivers a smaller channel count switch when upgraded to higher line rates. Switching of high-speed channels at line rates is highly desirable within DWDM networks, and is a feature that is reportedly not provided by the current generation of switches. A number of efforts are underway to develop switches with all-photonic cores (OOO). These switches would switch complete channels in a DWDM system providing low latency and scalability in a smaller footprint. However, they inherently offer only 1R functionality (amplification), no performance monitoring, no wavelength conversion and no capability for long-distance signal transmission. Also to be considered is the fact that the reliability of photonic technology remains unproven in the telecommunications industry. TeraBurst's patent-pending wavelength-level switching systems are based on an application of high-bandwidth optical-millimeter wave-optical or OMO technology. TeraBurst's OMO technology is capable of switching from low frequencies such as video signals around 6 MHz up to 40 Gbps (OC-768) today. The switch devices operate at full line rates (e.g OC-192 or OC-768) and render the switch fabric bit-rate and protocol-independent. The OMO technology also provides low latency and low power consumption. OMO technology enables TeraBurst's optical switching products to marry the signal quality and management capabilities of OEO with the speed and scalability of OOO technologies, at a price/performance ratio that exceeds both. On February 4th, the company announced its Wide Area Visualization Solution (WAVS), which is designed to provide connectivity and real-time collaboration capability between geographically dispersed locations of high bandwidth, 3D display and video applications, over optical networks. The solution combines the company’s high performance optical networking platform with recently developed digital imaging and audio/video/data mapping technologies capable of transporting video, audio and control data over a single wavelength in the optical network. “In building this particular product line up we actually ran across an application with wide area visualization solutions working in the visualization marketplace. We put together an interface that works within our switch and then working with supercomputer providers out there such as the SGIs, the Suns, the Hewlett-Packards, other computer manufacturers as well from that standpoint, to have graphic applications that will work in a visualization marketplace and we were able to take this information and go uncompressed over a lot of distances packed in a standard telecommunications packet which is a SONET-based OC-48 package and then deliver it in either a private or public network,” Said TeraBurst COO Tom Myers. “The advantage of this is the fact that we can now do global collaboration between multiple locations. And with the switch we have we can tie in multiple screens, multiple locations, allow users to pick and choose what they want to work with and then bring it down into their visualization location.” Applications for this technology include collaboration between different groups within the energy and government sectors, high-definition video distribution and computer-aided design collaboration. Initally, TeraBurst will focus on a 3-D stereographic video distribution application used in immersive visualization centers in the oil and gas industry. WAVS' ability to deliver secure, high-bandwidth, real-time connectivity across wide area networks enables users of high-end applications the opportunity to increase productivity and reduce resource constraints. TeraBurst recently tested the WAVS solution at the Center for Immersive Telecommunications for Global Exchange housed at San Diego State University (SDSU). “Well what happened is when SDSU was opening up their new building they put together this set-up for immersive visualization there and we set up the system to do a test at that location to demonstrate how this technology would work. The test went off great,” Myers commented. “We were actually able to do a full three-dimensional image at 98Hz which is 98 frames per second. You do it that way so when you put on some stereographic glasses you have 49 frames per second per eye which is a lot higher resolution than you’d get with the highest definition HD video. We’re able to do a loop back with that particular application to be able to show the data sent out, sent to an optical network, brought back in and received within this visualization center. “What we’re doing right now is we’re sending it up to be able to tie in with the California Institute of Telecommunications and Information Technology (Cal-IT2) which is based at the University of California at San Diego, and they’ll be doing collaborative work with each other. We’ll be setting that up in mid February to where we can do a number of different applications in a private network over a distance…actually a public network over a distance. So these groups can now do collaborative work back and forth. This is helping us get the demonstrations out to our customers so they can look at how the applications work and they can work in regional and global collaboration with this type of interface,” Myers concluded. Plans are also in place to link SDSU, the San Diego Supercomputer Center and Scripps Institution of Oceanography (SIO) over high-bandwidth fiber-optic connections. The collaborative aspects of the WAVS solution bring to mind the extremely hot topic of grid computing. One application that jumped to my mind right away was telemedicine. “Right now you can do telemedicine over video lines and you can make things happen from there but what we can do is take the computer images and be able to have a computer image stored in one location and then for example, an MRI or a three dimensional MRI stored in one location and then another location can take the image up, rotate it around, slice it up without bringing the data to that location,” Myers said. “The nice thing about this is the huge amount of data, the data files are giant. Also people don’t want to lose their data files, they want…the security aspects are in place, but they say, ‘we want you to be able to view it and work with it in a collaborative aspect, but we don’t want to give you the data because we own that data’ so another doctor or medical group can then work with it and consult back in forth without the data being transferred.” A somewhat obvious, but very positive factor of the telemedicine scenario Myers described and of grid computing in general of course is the elimination of both the transfer of massive amounts of data over the network and the resulting copies and storage issues that are born of that transfer. Twenty Terabytes doesn’t move quickly across anybody’s network and most people likely don’t want 20TBs of data sitting around in their warehouse even if it were to move quickly across a network. Time will tell whether TeraBurst is successful in the long run and I’ll leave any predictions to the experts. It is an interesting company though and their switches are positively smokin’. Also they seem well positioned, with both the switches and their visualization offering, to reap the rewards of the grid computing trend we’re experiencing now and the grid computing boom we’ll see in the near future.