Thomas Hawk, head of IBM's grid computing programme, defends and explains the technology concept and its goal to allow organisations to treat all their hardware and applications as one system or resource.
Hawk said he believed that kind of development was already taking place, adding that virtualisation is one example of an early form of grid computing.
Your vision of grid computing as a universal virtualisation layer for a broad range of enterprise IT services casts such a wide net, and strings together so many technologies that already do not work well, that it's hard not be sceptical about it. How do you defend it?
To me, it's all about the open-standards integration and acceptance. If we don't gravitate to open standards, then we will struggle. Companies buying technology have been clamouring for this for a while. They're candidly mad as hell and don't want to take it any more, and it's being driven by a level of complexity that they can't deal with.
The number one inhibitor to growing capability primarily from the IT function is the inability to manage complexity. So we see the virtualisation of the infrastructure over the appearance of open standards, and then the ultimate transformations of products and services to those standards, as a way to mask that complexity, reduce it significantly and increase manageability.
Are you working with your competitors on standards?
Yes. IBM, Sun, HP, Microsoft and BEA are stepping up more - all of those players are members of the Global Grid Forum to one degree or another. We are doing that collaboratively as a byproduct of the activities of the Global Grid Forum. But there are specific customer implementations that are also driving suppliers together.
Should users be pushing suppliers to ensure products have grid standards?
Absolutely. It is fascinating to me how little power the aggregation of the customer base believes they have. They have all of the power. Customers don't realise the clout and the power they have.
At what point does grid get beyond scientific and engineering applications and becomes used widely by businesses?
The thing that takes it to the next step is transaction management, where we can now begin to deal with, in a gridded infrastructure, the combination of traditional batch, job or compute-intensive workloads along with robust, transaction-oriented applications.
A lot of companies are heavily into virtualisation and increasing their utilisation rates for IT resources such as storage and compute power. If they are already accomplishing that, why bother with grid computing?
If they are accomplishing that... then they are doing grid. If they are doing virtualisation, it is grid. Maybe it's an early instance of grid. People who are virtualising their infrastructures - that's what they are doing. They are grid-enabling their capabilities.
The biggest grid advocates are the suppliers. When we talk to people out in the field, no one is saying they need grid systems to solve their broader business computing problems; it doesn't appear to be on many radar screens. Are we wrong?
I think a lot of the virtualisation activities are taking place. I think the exploitation of the infrastructures across heterogenous platforms is taking place. Whether or not people are calling that grid - some are, some aren't. But when you talk about the client base that we have in the Fortune 2,000, there's a pretty healthy understanding of and knowledge of grid.
Is there more to be done?
Yes. But I think the knowledge of the deployment and concepts is happening more rapidly.
Patrick Thibodeau and Robert L Mitchell writes for Computerworld
This was first published in July 2003