For this a new model for such high-end use is emerging - projects are already running for breast cancer research and linking universities in the United Kingdom and Netherlands - that aims to use large arrays of computers, and the spare capacity of others, to generate enough computing power for such medical discoveries and other research.

Called grid computing, it's a new turn from a company that has played a significant part in most of the major computer waves in the last few decades.

IBM were the original maker of the large mainframe computers that preceded the wave of personal computers (PCs) that are now the de facto computing platforms for most home and business users. They also made the original PC, which celebrated its 20th anniversary last year, using a then revolutionary idea to build it with off-the-shelf parts, and licensed its manufacture, which in no small part - along with a little software company called Microsoft -- led to the widespread usage we see today.

Now IBM is looking ahead and believes grid computing is "the next big thing" since the Internet.

"I've often been asked, 'What's the next big thing for the Internet?'" IBM's vice president for Internet strategies John Patrick said last year. "Until now, I didn't have the answer. I'm very confident now that the next big thing will be grid computing."

So, what is grid computing?

"A new open-source-based model of computing, grids are clusters of servers joined together over the Internet, using standard protocols and other open technologies, including Linux. Just as the world wide web allows people to share content via standard Internet protocols, grid computing allows widely-dispersed organisations to share applications, data and resources using emerging grid protocols," IBM says.

The analogy of a normal computer is a useful way to explain grids, says Hans Ulrich Maerski, chairman of the board for IBM's Europe, Middle East and Africa (EMEA) region.

The computer as we know it is a box housing a variety of components - the processor, memory, storage, etc - but in the grid computing model, and broadband connections, there is nothing to stop these components being in different parts of the building, country or even world.

"If you would say I'm going to put an ultra-fast, broadband capability between those things - the memory is in South Africa , the processor in Washington, the storage in Seattle - and I could run this with a standard OS and it doesn't make a difference to the performance, then we would be talking about the vision of grid computing," says Maerski.

The connections between all these elements will be the Internet, says IBM country general manager Mark Harris. "It is a new way to use the Internet as a top to bottom computing platform, integrating applications, middleware, servers, storage and services in an open, secure and cost effective infrastructure that can deliver computation in the same way that an electric utility can deliver power."

Indeed, increased computing requirements are pushing many companies to host their processing and storage needs in server rooms and data centres run by computer companies, including IBM.

Grid computing aims to link and harness the massive combined computing power of all of IBM's worldwide data centres to provide customers with the computing resources they need on demand, could help businesses reduce IT costs while expanding capabilities.

The backdrop to grids is that the computing industry has progressed through various phases, says Harris.

"We've come through the e-commerce environment and are now in the e-business stage. This is linked to transformation of organisation's needs to integrate islands of operations and offering in order to unify them. Corporations are struggling to justify the cost of building an e-business infrastructure," he says.

Harris says many companies will initially be seeking "storage on demand", that is "they don't have to make the investment in order to store, we will do it on your behalf". They will also need many other services on demand.

"IBM believes that grid computing and its protocols - like Linux and the Internet itself - will grow beyond the academic world and become a basis for the delivery of computing as a utility-like service over the Internet known as e-sourcing."

This e-sourcing or utility phase is the first step towards grid computing, they say.

"Grid computing will enable flexible, efficient e-sourcing, that is, the ability to buy/rent (virtual) computing resources and services on demand over the Internet provided by gigantic e-utilities, paying for what you use much as you would with electricity."

Across the industry computers and their service providers are increasingly being compared to a utility such as an electricity or water supplier, that must be available and working 24-7 (24 hours, seven days a week).

Echoing this theme about reliability, stability and what Microsoft calls "trustworthiness", Maerski says: "The first step have to make is to start making our computers - that we have in data centres - more available. It sounds strange because their availability is 99.99% but if you translate that into minutes of outage per years. That's quite few minutes."

Indeed, looking at it this way, there is a lot of capacity in such small increments accruing in server rooms and data centres.

All of this can be used collectively for the big number crunching needed in academic and pharmaceutical research. There's also an affordability issue, says Harris, as not every institution can manage to pay for the capacity they require nor can they drive the high volumes of processing and data on what they have.

"I don't think we've got much choice if we're going to play in certain markets," says Harris. "Things like research and education, especially at the university level, where an institution needs deep levels of computing and storage but will find it difficult to fund across multiple institutions, they will go that way."

"Grid computing is emerging first in the scientific and technical community, which is using [it] to enable scientists and engineers to collaborate in applications across institutions and around the world, such as high energy physics, life sciences and engineering design."

IBM is running one of the first such projects is with the University of Pennsylvania in the United States to perform remote breast cancer diagnosis and screening - digital versions of patient X-rays can be stored at remote sites and accessed instantaneously from the various medical centres.

Last year the British government awarded the contract to build one of nine grid centres in the United Kingdom, a high-tech data storage facility at Oxford University, to IBM as part of the government's plans to build a "national Grid" for collaborative scientific research. IBM was also chosen to build a similar grid in the Netherlands, connecting five universities.

Partly to demonstrate the feasibility of the technology, IBM's research department built its own grid: a geographically distributed supercomputer linking its research development labs in the US, Israel, Switzerland and Japan.

They are not the only ones working on this. Sun Microsystems announced their own Grid Engine Project last year, which includes software that locates idle computing resources, matches them to individual jobs and brings the computing power back to the desktop.

Grids at first glance appear to be similar in concept to other distributed computing models, but it differs in that it would let multiple users work simultaneously instead on just one task.

The most well-known example was the SETI@home scientific project, which used millions of individual desktop PCs during the night to help search for extraterrestrial life. Each computer processed a packet of data and returned its completed work to the central machine when it finished. This kind of peer-to-peer network became a popular idea and was behind the file-swopping music phenomenon that became with Napster, the MP3 sharing software that allowed music to be pooled between PC users for free.

Grid computing however links much more powerful resources: supercomputers, applications, servers and storage technology. Ultimately it will link up with another future technology: self-managing servers, towards which Project eLiza is working.

Maerski thinks grid computing doesn't differ from another much-publicised emerging technology, web services. These are Internet-based means to access personal information online or through a variety of wireless handheld devices - such as your contacts or diary - or pay for goods using a secure digital identity that knows you are you despite the device or access means you use.

"Web services are buying a service over web. It could be access to a library or transaction. And we will pay per transaction. I don't think it differs."

For companies, grids make sense because they don't need to have their own computing centre, IT staff, or pay for software licences. "We can provide you with all of this, They can be per user or per usage - they only pay for when they use it."

All of this is dependent on broadband - superfast data lines that are used to connect big businesses to their satellite offices. In Europe and the United States, where demand makes it affordable, they are available by ordinary consumers to connect to their Internet service provider (ISP).

The cabling required for grid computing is optical, which is the most cost effective and effective way to do it, says Harris. "Most of the projects I've seen that's they way they have done it."

But this is a problem for South Africa, where broadband on a consumer level is scarce and is still pricey for corporates. Our infrastructure is thought to be five years away from being able to hook into an international grid such as the ones being built now.

However as Harris points out in Ghana optical cables are being laid at the same time as electrical cables.

"They are already saying we have to hook our universities together and share resources. That is there in many of the African countries and I don't think SA will be too far behind."

                                                                                                          Back to story list >>




                                                                           [content is still king]    [about us]    [clients]    [contact us]