LambdaGrids para e-Science

The world of distributed computing and grids continues to evolve. First we had parallel computing, then distributed computing, then grids and now lambda grids. The Optiputer, TeraGrid and The Netherlands DAS-3 are good examples of these new types computational facilities where the optical network is an integral part of the computational infrastructure as opposed to a piece of plumbing. The DAS-3 for example will have 8 lambdas riding on top of the SURFnet network interconnected to clusters located around the Netherlands. Recently an Optiputer node was established at Communications Research Center using the new lightpath connection between CENIC and CA*net 4. The following are some excerpts from GridToday -- BSA

www.gridtoday.com
www.teragrid.org
www.optiputer.net
http://www.gigaport.nl/info/toepassingen/lichtpaden/das.jsp

The OptIPuter: 21st Century E-Science
The OptIPuter project -- named for its use of optical networking, computer storage, processing and visualization technologies is a 21st-century prototype cyberinfrastructure that tightly couples computational resources over parallel optical networks using the internet protocol (IP) communication mechanism.

The OptIPuter exploits a new world in which the central architectural element is optical networking, not computers, said project manager Maxine Brown speaking at the first annual TeraGrid conference in Indianapolis.

"The goal of this new architecture is to enable scientists who are generating terabytes and petabytes of data to interactively visualize, analyze and correlate their data from multiple remote storage sites connected to optical networks," said Brown, associate director of the Electronic Visualization Laboratory at the University of Illinois at Chicago.

"However, this time the parallelism is in multiple wavelengths of light, or lambdas, on single optical fibers, creating supernetworks."

Researchers at the University of Illinois at Chicago and the University of California at San Diego are leading the OptIPuter project effort under a five-year National Science Foundation (NSF) Information Technology Research (ITR) grant, funded for the period October 2002 through September 2007.

According to Brown, the OptIPuter's mission is to enable collaborating scientists to interactively explore massive amounts of previously uncorrelated data. The researchers on the project hope that the OptIPuter, when linked with remote "data generators," whether the TeraGrid, instrumentation or data storage devices, will prove to be an enabling technology for large-scale networked science facilities, as well as for broader societal needs, including emergency response, homeland security, health services and science education.

The OptIPuter project, said Brown, is different from other distributed Grid and high-performance computing projects in that it focuses on
optical networking.

"Metro and long-haul 10Gbps optical networks are 100 times faster than 100T-base Fast Ethernet local area networks connecting PCs in research
laboratories," Brown said. "The exponential growth rate in bandwidth capacity over the past 12 years has surpassed even Moore's Law due, in
part, to the use of parallelism in network architectures. Now the parallelism is in multiple wavelengths of light on single-strand optical fibers, creating supernetworks, or networks faster (and someday cheaper) than the computers attached to them."

While extremely important, bandwidth alone is not the solution, Brown continued.

"The OptIPuter is working on new Grid computing paradigms -- new middleware, transport protocols, optical signaling, and control and management software to enable applications to dynamically manage lambda resources just as they do any grid resource, creating a 'LambdaGrid' of interconnected high-performance computers, data storage devices and instrumentation."

The OptIPuter project is learning to take advantage of bandwidth and storage to conserve "scarce" computing in today's new world of inverted values. Essentially, the OptIPuter is a virtual parallel computer in which the individual processors are distributed clusters; the memory is large distributed data repositories; peripherals are very-large scientific instruments, visualization displays and/or sensor arrays; and the motherboard uses standard IP delivered over multiple dedicated lambdas that serve as the system bus or backplane.

"CAVEwave" is a dedicated 10-Gigabit deterministic network that is being deployed nationally over the National LambaRail (NLR) for use
for OptIPuter experiments. In turn, NLR is connected to an international fabric of optical networks, which are part of the Global Lambda Integrated Facility (GLIF). GLIF connects collaborators to their colleagues and to data sources all over the world, providing researchers with guaranteed bandwidth for data movement, guaranteed latency for visualization/collaboration and data analysis, and guaranteed scheduling for remote instrument control.

"The OptIPuter project is not optimizing toward scaling to millions of sites, a requirement for commercial profit, but empowering networking
at a much higher level of data volume, accuracy and timeliness for several key high-priority research and education sites."

To learn more about the OptIPuter project, visit: http://www.optiputer.net/.