Articles

When the network is as fast as the computer's internal links, the machine disintegrates across the net into a set of special purpose appliances.
-- Gilder Technology Report, June 2000

India supercomputing efforts

Once stymied by US sanctions but now aided by fast communications links, Centre for Development of Advanced Computing State-run agency for advanced computing plans to build a nationwide grid of supercomputers for mammoth applications. The grid would share or combine diverse computer memories and software in parallel processes to aid environmental modeling, fast analysis of satellite images, advanced chip design and simulation of heavy-duty equipment like turbines.

C-DAC plans to link the seven prestigious Indian institutes of technology (IITs), the Bangalore-based Indian Institute of Science and other academic institutions in the I-Grid, Arora said. The building of the grid would coincide with India's 10th five-year development plan, starting in April, but the financial details have not been worked out yet, he said.

Founded in 1988, C-DAC faced more than a decade of technology export restrictions by the United States, particularly relating to the Cray supercomputer, on the grounds that India might put the technology to military use. Braving the sanctions, C-DAC has built four versions of its Param series of machines, putting India in an elite club of supercomputing nations like the United States, Japan, Israel and China.

The latest Param crunches numbers at a speed of 100 gigaflops, or 100 billion floating point operations per second. That put its among the world's high performers, although the United States has developed machines 10 times faster, a feat C-DAC is trying to match.

n the 1980s, the National Science Foundation created the NSFnet: a communications network intended to give scientific researchers easy access to its new supercomputer centers. Very quickly, one smaller network after another linked in - heralding the Internet as we know it today.

In 2002, the same National Science Foundation will install hardware for the TeraGrid, a transcontinental supercomputer that should do for computing power what the Internet did for documents. First, clusters of high-end microcomputers will be set up at four sites: the National Center for Supercomputing Applications at the University of Illinois at Urbana-Champaign; the US Department of Energy's Argonne National Laboratory outside Chicago; Caltech in Pasadena and the San Diego Supercomputer Center at the University of California, San Diego. Then, by early next year, those four clusters will be networked together so tightly that they will behave as a single entity. This virtual computer will rip through problems at up to 13.6 trillion floating-point operations per second, or teraflops - eight times faster than the most powerful academic supercomputer available today. The TeraGrid is a prime example of what has come to be known as "grid computing" - the massive integration of far flung computers, databases, and scientific instruments over the Internet or a virtual private network and promises IT power on demand.

The grid - a kind of hyper-network that links computers and data storage - is owned by different groups so that that they can share computing power. Computing grids get their name from their similarity to power companies' electrical grids. When one turns on a light, one's not necessarily grabbing electricity from the nearest power station but rather tapping into a regional grid of stations. If the station the light is drawing power from goes off-line, the grid simply routes electricity from another plant. In much the same way, a computing grid gives access to networked computing resources. The grid ties together a group of machines, letting one borrow their unused or underutilized resources regardless of make, model, hardware configuration, or sometimes even operating system. Grids can tap into dedicated client PCs, employee workstations, servers, or mainframes. Tasks may run solely on one machine, or run on one, then move to another as processor cycles become available. Or they may run on multiple machines simultaneously if they can be separated into pieces. For example, if an enterprise and own lots of servers and wants to share the computing power of all those servers, and, most important, wants to share the data and the applications in all those servers, with them all integrated on a single network, all the people that the enterprise is working with can share those same resources. Today, computers spend much of their time sitting idle, as a processor waits for data to crunch. In a grid world, the idle time of hundreds or thousands of servers could be harnessed and rented out to anyone who needed a massive infusion of processing power. A large bank's computing resources may be largely unused overnight. If connected to a grid, that dormant power could be tapped by an unrelated company on the other side of the globe for a variety of tasks. In return, the bank might be given access to extra capacity during its peak hours.

The use of an intelligent distributed network in which computing power becomes a resource and a service, as an electric utility, is not confined to hypothetical reality. The academic community, long fascinated with the potential to link together the world's computers to tackle massive computational and social problems, such as forecasting weather and curing intractable diseases, has been the forerunner in adopting grid computing. University of Pensylvania, for instance, in its research for breast cancer, exploited IBM's grid computing systems, a visionary patient centric medical record system that could capture from any location the full range of health care files including high fidelity patient medical images records and clinical history. Grid technology has largely been limited to the life sciences and pharmaceutical companies for applications such as genome sequencing, simulations and data analysis, and collaborative engineering. In the industry the grid computing space can be divided into three categories: infrastructure suppliers, software, and niche companies. Among infrastructure suppliers, IBM and Sun Microsystems have been "very aggressive," and Hewlett-Packard and Compaq are making a push. Microsoft, Intel and systems integrators are also focusing on Grid computing. Niche companies include Platform, Avaki, Entropia, United Devices, Terrphpring and Juno. Microsoft is also developing grid-computing software for use with its products, and the software giant has decided to back one of the key grid organisations, the Globus Project, whose Globus Toolkit software lets organisations set up grids and manage grid jobs. Microsoft is funding Globus, a research and development project focused on enabling the application of Grid concepts to scientific and engineering computing is being backed By IBM and Microsoft currently, with $1 million to make sure the toolkit runs on Windows XP and on its .Net Web services software infrastructure. Other baxkers include companies like IBM.

The benefits for the future are compelling: One, within the organisation; grid computing allows corporate users to treat a company's entire IT infrastructure as one computer, grabbing unused resources as they are needed. According to a Gartner study, Intel servers only operate at 5 per cent to 20 per cent of capacity during the workday - and more or less zero at night. Grid computing will ensure that computing power of these machines can be tapped. Two, the concept allows widely dispersed organisations to form virtual organisations that collaborate on common problems by enabling them to share everything from engineering blueprints to software applications. IBM's Dr. Irving Wladawsky-Berger is reported to have said in a grid computing conference, "In fact, the community that is developing Grid computing thinks its biggest impact will be the ability to develop virtual IBM's." Three, lower the total cost of computing by enabling the sharing, efficient optimization and overall management of those computing resources. Four, grid computing would lead to the emergence of new business models focused on outsourcing, utility computing, peak load support, and capacity and disaster planning. Probably it chief appeal at the moment is that grid comuting would ensure business continuity.

Despite these benefits grid computing has a long way to go. The grid vision is "resource sharing and coordinated problem solving in dynamic, multi-institutional virtual organizations," providing on demand, ubiquitous access to computing, data, and services, with new capabilities constructed dynamically and transparently from distributed services. Grids are built from multiple vendor components, so interoperability of standards, , just as the Internet Protocol (TCP-IP) is at the heart of the Internetv, is imperative. No formal standards process as yet exists for Grids.

Globus is an organization seeking to set standards for grid computing similar to the standards that were set up to run the World Wide Web. One of the more significant standards proposals at Global Grid Forum 4 was the Open Grid Services Architecture, a method of marrying Web services technology to grid technology The OGSA is based on a four-layer model: at the top are user applications; then comes what is known as collective services, which includes directory handling, diagnostics and monitoring; below that, resource and connectivity protocols handle access to servers and networks; and finally comes the Fabric, which is everything on the network--storage, computers, connections, sensors and so on. The fabric and the user applications are familiar ground: it's the middle two layers that define the Grid. But considering the hetorgenity, specific consensual standards are still evolving and much of the work can be clubbed "work in progress." Also security remains a constant problem, delaying acceptance of external Grids. For now, analysts such as IDC believe that most adopters are taking the "intraprise" approach, staying within the enterprise instead of going beyond the firewall to the Internet.

Despite the promise of grid computing, the truth is that its has not wanderd far from academic and government conclaves. Considering the rapid progress being made in another decade grid computing will soon be the accepted scientific reality of the 21st century.