The internet changed the way companies and individuals thought about information and content, and established the notions of open standards and collaboration.

Now "grid computing" promises to do the same for IT systems and applications - tapping into the enormous reserves of under-used computer power in organisations to deliver cost-effective computer resources on demand.

Definitions vary somewhat, but Platform Computing, one of the pioneers of the technology, defines grid computing as "the co-ordinated, transparent and secure sharing of IT resources across geographically distributed sites."

Like the web, the origins of grid computing can be traced back to the early 1990s when the scientific community began looking for ways to make better use of existing computer resources, to share applications and to collaborate on projects through distributed computing.

However, in the last year or so the concept of grid computing has begun to enter the IT and commercial mainstream with all the leading enterprise computing vendors including IBM, Hewlett-Packard and Sun Microsystems embracing it as a core component of their competitive strategies.

There is also broad agreement that grid computing represents the next wave of enterprise computing and will have profound implications for the both IT systems vendors and users.

"Grid computing will take e-business to the next level by giving customers a resilient, flexible, virtual IT infrastructure readily available from any location, on demand," says Irving Wladawsky-Berger, vice president for technology and strategy in IBM's server group. "By grid-enabling our products we will give customers the ability to share computing resources, such as applications, data and computing power, both internally over intranets and externally over the internet."

Resources

Advocates argue that grid computing will enable enterprises to more effectively manage, pool and share resources including computers, storage and software applications.

It will enable them to be used as a single, unified and managed resource within a department, an enterprise or, using the internet, between businesses.

What is more, they foresee an environment in which companies and individuals will access computing power and resources such as storage, software applications and services with the ease we associate today with pushing a plug into an electrical power socket.

In the world of grid computing, IT resources are a utility delivered like electricity, water and the telephone dial tone -a seamless infrastructure that will support next generation collaborative working, virtual super-computing and automated web services.

In fact the basic idea behind grid computing is relatively simple - take under-utilised IT resources, such as PCs, workstations and storage devices, add a management software layer that can pool, manage and allocate these resources dynamically based on priority rules, and then enable users to call upon these shared resources as they need them.

Grid software - often called a resources manager - provides the "glue" for grid networks ensuring that all requests for resources are managed centrally. Individual applications do not need to know where these resources physically are.

Typically IT vendors such as including Platform Computing envisage three evolutionary types of grid:

* Enterprise grids: These enable collaboration and sharing of computer resources across multiple sites and geographic locations within and between companies.

* Partner grids: These are an extension of enterprise grids that enable transparent, secure and co-ordinated resource sharing and collaboration among partner organisations.

* Service (or global) grids: A single, internet-powered service grid will provide computing power "on-tap" for individual access around the globe, creating one virtual computer built on open protocols with everything shared.

Most of the grids built so far serve the scientific and technical community. For example, IBM's customers include the UK National Grid, the Dutch Tera-Grid Facility, the University of Pennsylvania Mammography Grid, and the North Carolina Bioinformatics Grid.

Outside academic and research establishments a few leading edge companies have deployed enterprise grids, usually seeking to exploit the potential for collaborative design or rapid prototyping.

Among the early adopters are AMD, the microprocessor chip group, which is using grid computing to extend the barriers of electronic design. The life sciences industry is also using grid computing techniques to speed the mapping of the human genome and to accelerate the time to market for important drugs. All four of the major genome sequencing centres use Platform's grid technology.

But for most enterprises, the immediate appeal of grid computing is economic. As Rick Hayes-Roth, chief technical officer for Hewlett- Packard's software organisation, notes: "There are plentiful computing resources, but most of the time these resources are idle."

It is estimated that only 5 per cent of the capacity of PCs and 20 per cent of servers deployed by enterprises are used. An enterprise-wide grid can raise this usage level to 80 or 90 per cent, claims Wolfgang Gentzsch, the director of grid computing at Sun, saving millions of dollars, speeding research, time to discovery and ultimately time to market.

Mr Gentzsch, who joined Sun in July 2000 with Sun's acquisition of Gridware, a US and Germany-based software company, believes companies will use these additional resources, for example, to speed products to market or undertake more product testing.

Grid computing advocates also argue that a number of other factors have come together to create a favourable climate for the adoption of grid computing outside the laboratory. Among these, the growth of the commercial internet has led to the widespread deployment of internet infrastructure and acceptance of the need for open-systems standards.

Meanwhile, high profile projects such as seti@home (The Search for Extra Terrestrial Intelligence), Decrypthon in France (search for a cure to cancer) and Stephen Hawking's Cosmos project (mapping the development of the galaxy) all use grid technology.

They have also highlighted the potential of distributed computing systems to divide large computational problems into small chunks that can then be shared out among thousands of geographically dispersed machines.

Finally, the trend towards service-centric computing model and the delivery of computer resources via web services (see articles on Pages 1-6) is pushing organisations towards a transparent grid computing framework, says HP's Mr Hayes-Roth.

The IT industry has also learned from the development of Linux, the open source operating system, and web services that it pays to co-operate on defining open standards. Open protocols are essential to grids because they enable heterogeneous systems to work together as a single system.

New initiative

The drive for grid computing standards has been led by the Globus open source development community, the New Productivity Initiative (NPi) and the Global Grid Forum. Last month the NPi and Global Grid Forum agreed to join forces to create the largest global grid computing standards group.

By consolidating their efforts, the two groups hope to stimulate even broader industry participation and collaboration on the common standards required to drive adoption of grid computing and enable it to deliver on it promise.

Robert Gordon, Platform Computing's chief executive notes: "Grid computing has been described as the fifth wave of the IT revolution. If the first wave was the mainframe computer, the second was the desktop PC, the third wave the client server, the fourth wave the internet, the fifth wave is distributed computing.

"The third wave put information technology at everyone's fingertips, and the fourth wave gave everyone access to information anywhere in the world. The fifth wave will give individuals and organisations access to unlimited computing power."

EMAIL THIS PRINT THIS MOST POPULAR