Taking the simple concept of strength in numbers and applying it to tap the unutilised processing power of networked PCs has brought about a revolutionary concept called Grid Computing. Srikanth R P elaborates on the benefits and the far-reaching effect this concept could have on the Indian processing scenario

Think about this: How comfortable would you be if you were asked to have your own electricity generator to generate electricity? You would obviously find the concept bizarre. But years ago, Ford Motor Company had its own power plants to generate electricity. However, when supply of electricity became more dependable, Ford sold off these plants and got out of the utility business. This line of thinking is now being adopted by global giants like IBM and Sun who are pushing forward the concept of ‘Grid Computing’. In simple terms, it means that a network of computers, connected together via the Internet, satellite or optic fibre links can be used by various organisations for sharing processing power—translating the thousands of computers with idle CPU power into a single virtual supercomputer. Processing power will be available at the switch of a button, just like electricity is available today.

R K Arora, executive director, C-DAC, says it is proposing to build a grid which will connect around 15 supercomputers

Hence, if organisations don’t have billions of dollars to invest in hardware for highly data intensive processes like bioinformatics, aeronautical design, climate modelling or drug research—they can still get the processing power of a thousand PCs or more, using the grid computing concept. A large company could also use a grid to connect geographically diverse offices or unify its supply chain. Also possible is the creation of a virtual database that can pool content from different offices and take on large engineering and modelling projects, which are beyond the scope of a single computer.

All a customer needs is access to the grid via the Internet or any other link, and grid software, which distributes pieces of a program to thousands of computers for processing. One of the most famous and publicised examples of Grid Computing is the SETI@Home (Search for Extraterrestrial life) project in which PC users all over the world donate unused processor cycles to help in the search for signs of extraterrestrial life by analysing signals coming from outer space. Another ambitious project is one undertaken by Oxford University’s Centre for Computational Drug Discovery that utilises more than one million PCs to look for a cure for cancer. People around the world donate a few CPU cycles from their PCs through screensaver time. The project will eventually analyse 3.5 billion molecules with cancer-fighting potential. So far, more than 50,000 years of CPU power (based on a 1.5 GHz chip) have been put to work.

Taking the grid to corporates
While Grid Computing primarily started in academic institutions, giants like IBM and Sun are looking to give it a corporate twist with their initiatives. For instance, globally IBM is looking to create a global arrangement through which companies, instead of buying a set amount of processing power that often goes unused, will now be able to opt for a pay-per-use facility by dipping into a vast pool of computing resources.

Just the way it played a pivotal role in pushing the commercial adoption of Linux, IBM is also one of the frontrunners in popularising the grid concept. It is looking to push the concept of Grid Computing by working with the Globus open source development community and the industry standards body, Global Grid Forum. Open protocols are essential to grids as they enable heterogeneous systems to work as a single system.

Though Grid Computing is still very nascent as a concept, there are some interesting developments, which could see the concept gain ground.(See Box: Other global Grid Computing projects) For example, IBM is looking at bringing together the Web services and Grid Computing efforts. This would create, for the first time, a way to use the Internet as a top to bottom computing platform, integrating everything from the application to the server. IBM has already announced a set of specifications that would unite these two concepts.

IBM’s M Ganesh, says that Grid Services have emerged as an important new field in computing

India’s own grid—IGrid
The Grid Computing concept is not restricted to the global scenario, and companies like IBM are looking to extend their expertise in the Indian sub-continent too. Says M Ganesh, country manager, Enterprise Systems group, IBM India, “Grid services have emerged as an important new field in computing, with its focus on enabling flexible, secure, interoperable, and co-ordinated resource sharing among dynamic collections of institutions and resources. IBM’s entire server portfolio is defined for Grid Computing capability. All of them have IBM’s Project eLiza capability–which allows self-managing, self-healing and self-protecting capabilities–and this is a significant aspect in the area of Grid Computing. We are specifically exploring opportunities in India for providing high-end computing solutions across all platforms for sectors like Life Sciences, especially biotechnology and bioinformatics, using the concept of Grid Computing.” Though this concept will be initially targeted at academic institutions that need a tremendous amount of processing power for their research-based activities, the trend could well be extended to commercial organisations, if the concept catches on.

Also noteworthy are C-DAC’s ambitious efforts to make Grid Computing a reality. While C-DAC has achieved significant milestones in Indian history, such as making India’s first supercomputer, it admits that the full potential of supercomputers have never been utilised. C-DAC is proposing to build a grid, which will connect around 15 supercomputers at an initial investment of Rs 120 crore. The project, christened ‘IGrid’, is India’s first Grid Computing project and will provide computing power of 1,000 gigaflops upwards. This, industry analysts believe, is a logical extension to the work in the area of High Performance Computing and Communications (HPCC), which C-DAC has been engaged in for over a decade. This is also in line with the global trend in this area and is rapidly being seen as an effective method of providing seamless access to supercomputing facilities to a large number of researchers rather than having to own them individually.

Agrees R K Arora, executive director, C-DAC, “In the Indian context, though HPCC has received national level attention and is seen as a strategic resource, the fact is that the need for high-end computational resources are met only by select organisations. Hence, there is a growing need for access to high-end computational resources to serve the interests of a large number of scientific and engineering resources and institutions countrywide. The grid-enabled computing infrastructure is an ideal candidate, as it will allow cost-effective access to high-end computing resources rather than individual institutions having to own the same.”

The proposed IGrid will consist of supercomputing infrastructure across eight major cities, tentatively earmarked to be linked with an aggregate computing power of 10 teraflops. C-DAC has proposed the grid in such a way that organisations across cities are prime candidates for using processing power equal to supercomputers. Spanning over two phases, the first of which will link C-DAC’s centres at Pune and Bangalore to prove the concept of the grid, the second phase will connect the remaining cities. The proposed grid will have at least 100 terabytes of storage capacity to help compute and process data. The link– up of the supercomputers would take place through high bandwidth optic fibre connectivity offered by BSNL or other basic service providers.

Grid’s key drivers
Cost is a prime factor that drives organisations to give priority to new and emerging technologies. At a time when IT budgets are shrinking, the concept of Grid Computing could prove to be beneficial for CIOs and CTOs. Says Ganesh, “The promise of Grid Computing will cause a shift from the sharing of information enabled by the Internet, to a new level that will allow the sharing of computing resources. At a time of limited budgets and growing computing requirements, the time is ripe for Grid Computing to take its place as the technology trend for the new millennium. Organisations and institutions that engage in mission critical applications would benefit from Grid Computing initiatives. In addition, we see a lot of opportunity in government initiatives, educational institutions, scientific and technical organisations.”

Like other supporters of this concept, Ganesh too believes that Grid Computing and grid protocols like Linux and the Internet will grow beyond the academic world and become a basis for the delivery of computing like a utility.

Globally, the significant drivers for Grid Computing have been primarily in the areas of bioinformatics, climate modelling, computational fluid dynamics, space science and high–end storage services like digital libraries, e-learning and GIS–based spatial information services. The same trend could apply to our country too as all the applications are relevant and important to India.

In light of the above, C-DAC’s IGrid could prove to be beneficial as it proposes to offer a seamless, pervasive and inexpensive access to high-end computational resources by institutions across vast geographical areas.

Says Arora, “The proposed IGrid would be of use to both research and academic institutions and also to industries. For example, areas such as structural mechanics, fluid mechanics, storage services, digital 3D animation, chemicals and polymers, genomic sequencing and healthcare applications would be of tremendous interest to commercial organisations. The emerging field of bioinformatics covering genomic and proteonomic research would require large distributed computing infrastructure. Grid Computing lends itself as an effective solution to this, and this would drive a number of research and business interests.” In addition, small software companies can think of collaborating with other equally competent firms and develop products through a collaborative development, which can be enabled through a grid.

Besides the cost advantage, there are other important features, which can tilt the scale in favour of Grid Computing. For example, it allows features like resource aggregation, which allows corporate users to treat a company’s entire IT infrastructure as one computer, thereby allowing more efficient management.

Also supported are features like database sharing, which allows companies to access large remote databases. (This is particularly useful for Life Sciences firms, which need to share human genome data with other companies.) In addition, features like Collaboration allow widely dispersed organisations to work together on a project. Though Grid Computing is still in a nascent stage, there are compelling reasons which could pave the way for its adoption. Who knows, one day your PC’s processing power could be used to discover a new drug!