January 10, 2005
In December, 84 students enrolled in a graduate course on Information Technology & Public Policy sat down for their final class -- but not in the same classroom. Instead, the students were scattered among four separate locations, thousands of miles apart. Eighteen students took the class for credit at the University of California-San Diego (together with a dozen students auditing the course), with roughly 20 students each at the University of Washington (UW), UC-Berkeley and Microsoft Research.
"It didn't feel like 'being there,' but that did not necessarily detract from the experience," said Jonathan Weinberg, a second-year Master's candidate in computer science at UCSD. "The goal of this technology should not be to imitate the in-person experience, which it cannot, but rather to achieve something that is qualitatively comparable, which I think it did."
The far-flung course introduced grad students to various information technology (IT) policy issues, including intellectual property law and economics, technology transfer, workforce issues such as outsourcing, privacy and encryption, and more.
"A distributed classroom offers enormous advantages -- diverse faculty perspectives, diverse student perspectives, the ability to attract phenomenal guest speakers," said Ed Lazowska, a professor of computer science at UW. "The technology is not perfect, but it's much better than 'good enough,' and this course, while also not perfect, was far richer than any one institution could have provided on its own."
Lazowska was the principal instructor, along with Steve Maurer of UC Berkeley's Goldman School of Public Policy. In addition, guest speakers included Dave Dill (Stanford) on electronic voting; Rich Newton (Berkeley) on technology transfer; Bob Gomulkiewicz (UW Law School) on open source contracts; Ed Felten (Princeton) on spam, DMCA/DRM, and copyright issues related to peer-to-peer (P2P); Brad DeLong (Berkeley) on the economics of global sourcing; and Tap Parikh (UW) and Eric Brewer (UC Berkeley) on ICT and rural development.
Computer Science and Engineering (CSE) professors William Griswold and Geoff Voelker organized the course at UCSD. "Our students were able to benefit from the equivalent of a distinguished lecture series without the university having to budget $15,000 to fly everyone in," said Voelker. "Students also benefited because much of the material was recent and timely, so having the experts teach it via distance learning 'fast-forwards' the percolation process." He also noted the broader perspective that came from working with faculty and students from other universities as well as Microsoft engineers enrolled at UW.
The university classrooms were linked to each other and to Microsoft in Redmond, Wash., via Internet2 networking, with support from Microsoft's ConferenceXP software platform, which is the result of a collaborative research effort between Microsoft Research and a number of universities. It uses a P2P architecture and IP Multicast network topology to ensure scalability. "A single peer can join a ConferenceXP session without affecting the remaining clients," said Bryan Barnett, group manager for university relations at Microsoft Research. "Each client machine communicates directly with each of the other three. It's no longer a hub-and-spoke model." ConferenceXP also allows each location to transmit and receive at various bandwidths.
"Traditional distance learning has been delivered by disseminating a lecture from a single site, where the professor resides, to a number of remote locations," noted Griswold. "In this course, each site could see each of the other three sites and talk to them freely without worrying about any technological limitations."
The distributed course also used Classroom Presenter, a ConferenceXP-based application developed by UW Computer Science & Engineering Professor Richard Anderson, first while on sabbatical at Microsoft Research and subsequently at UW. Classroom Presenter is a distributed presentation system for Tablet PCs that provides "virtual mylar" -- the ability to "mark up" (and archive) PowerPoint transparencies. "Classroom Presenter breathes the life back into PowerPoint presentations," explained Lazowska. "It also provides great features for student interaction that we didn't utilize in this course, since only the instructors used Tablet PCs -- that's for next year!"
Apart from scalability, Microsoft's Barnett says the main goal of ConferenceXP is to close what he calls "the fidelity gap." "Most applications now deliver something that is roughly 75 percent faithful to the in-person experience," he said. "With Internet2 bandwidth, we can close the gap further in terms of very low latency, high-quality video with very good resolution, and multiple video and data streams being delivered simultaneously and independently."
That goal is shared by researchers on a number of high-speed networking projects underway at UCSD and the California Institute for Telecommunications and Information Technology [Cal-(IT)2], with which both Griswold and Voelker are affiliated. The National Science Foundation is investing nearly $19 million on three such projects: the OptIPuter; the Fast Wired and Wireless Grid (FWGrid); and the Laboratory for the Ocean Observatory Knowledge Integration Grid (LOOKING). LOOKING itself is a joint venture of UW, UCSD and other institutions, and is led by UW's Lazowska, with co-PI and Cal-(IT)2 director Larry Smarr, a computer science professor at UCSD. The OptIPuter and LOOKING both involve distributed scientific collaboration among researchers at UCSD, UW and other sites, over dedicated lightpaths that will outstrip the bandwidth now available over Internet2 connections.
"All of these projects are exploring the opportunities offered by fat-pipe networking," said Griswold. "We've been used to 1MB per second of bandwidth between any pair of computers, and now we're starting to ask what we can do if we go to 1GB or 10GB and beyond."
Microsoft's Barnett is the first to admit that current technology is not yet perfect. "We're pushing the envelope. IP Multicast is not widely supported by universities connected to Internet2," he said. "It also requires a great deal of care at each site to ensure that the best equipment is available and that the technology is working properly, in order for the feed coming from the classroom to be consistent and dependable."
Technology support during the IT & Public Policy course came from staffers at Microsoft and all three campuses, including Steve Hopper and Marvin McNett at UCSD, Rod Prieto and Fred Videon at UW, as well as UC Berkeley's Marvin Motley and Arthur Yeap, and Tim Chow at Microsoft.
"ConferenceXP is a research platform designed to facilitate various kinds of research," said Microsoft's Barnett. "This course was one of several demonstrations we are undertaking with various partners to gauge how close we are to fulfilling our early objectives."
The software giant's research arm will support future distance-learning projects, including a winter course that began on Jan. 4 on transaction processing for e-commerce. The principal instructor is Microsoft's Phil Bernstein, a pioneer in the field. Enrolled students are located at UW, Microsoft Research, and UCSD, where the faculty coordinator is CSE professor Yannis Papakonstantinou. Also scheduled: an international collaboration linking the Eisenberg School of Business at the University of Massachusetts with Ireland's National University in Galway.
Noted Barnett: "Without pre-judging the results of these demos, we do appear to be approaching the point where this type of distance learning becomes feasible."
Microsoft is also looking to adapt its software for tele-medicine and remote control of lab equipment -- areas where UCSD researchers are already active, notably as part of the Cal-(IT)2-funded Smart Vivarium project.
"Our job is to answer the question: What are the benefits to users if we can provide them with this quality of experience," concluded Barnett. "It may not be sufficient, but our job is to create those experiences and ask how the dynamics of education will change, and if so, whether we can close that fidelity gap [between remote and in-person classroom experiences] to within 5 percent."
"What makes this particular experiment valuable is that it points the way to a fundamental change in distance learning to a multi-point, peer-to-peer model," added Griswold. "These new fat pipes of bandwidth will change not only the equipment we use, but also the way we teach and learn."
"This course allowed us to take the classic classroom environment and then augment it with advantages that distance learning can provide over traditional techniques," summed up UCSD's Geoff Voelker. "We still need to address the difficulty of getting students to interact more easily and more often with students and instructors at other locations. But the bottom line is whether the benefits outweigh the shortcomings. For this course, the answer was yes, without a doubt."
Large-scale, worldwide scientific initiatives rely on some cloud-based system to both coordinate efforts and manage computational efforts at peak times that cannot be contained within the combined in-house HPC resources. Last week at Google I/O, Brookhaven National Lab’s Sergey Panitkin discussed the role of the Google Compute Engine in providing computational support to ATLAS, a detector of high-energy particles at the Large Hadron Collider (LHC).
Frank Ding, engineering analysis & technical computing manager at Simpson Strong-Tie, discussed the advantages of utilizing the cloud for occasional scientific computing, identified the obstacles to doing so, and proposed workarounds to some of those obstacles.
The private industry least likely to adopt public cloud services for data storage are financial institutions. Holding the most sensitive and heavily-regulated of data types, personal financial information, banks and similar institutions are mostly moving towards private cloud services – and doing so at great cost.
May 16, 2013 |
When it comes to cloud, long distances mean unacceptably high latencies. Researchers from the University of Bonn in Germany examined those latency issues of doing CFD modeling in the cloud by utilizing a common CFD and its utilization in HPC instance types including both CPU and GPU cores of Amazon EC2.
May 10, 2013 |
Australian visual effects company, Animal Logic, is considering a move to the public cloud.
May 10, 2013 |
Program provides cash awards up to $10,000 for the best open-source end-user applications deployed on 100G network.
May 08, 2013 |
For engineers looking to leverage high-performance computing, the accessibility of a cloud-based approach is a powerful draw, but there are costs that may not be readily apparent.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/02/2012 | AMD | Developers today are just beginning to explore the potential of heterogeneous computing, but the potential for this new paradigm is huge. This brief article reviews how the technology might impact a range of application development areas, including client experiences and cloud-based data management. As platforms like OpenCL continue to evolve, the benefits of heterogeneous computing will become even more accessible. Use this quick article to jump-start your own thinking on heterogeneous computing.