Information Age: News, analysis & insight for IT & business leaders

“If it’s a choice between buying software and adapting it, or buying adaptable software, I think the answer is pretty easy.” So says Dr John Manley, director of utility computing at HP Labs, the IT giant’s global R&D facilities. But he also sounds a note of caution: “It’s not really an answer for today, but one for tomorrow.”

As business leaders hunger for software that is more easily customisable to their specific needs, service-oriented architecture (SOA) is only one of the necessary first steps. In fact, the overall objective of being able to adopt adaptable software – and pay for it on the basis of usage – will ultimately require that software is delivered as a utility service. And the role of the service provider will be key.

Manley illustrates this theme through the example of the processing-intensive image rendering software used in the blockbuster animation features such as Madagascar. Five years ago, HP began offering film animators the option to lease time on a large grid of servers.

The pay-as-you-go models held considerable appeal. Equipped with the HP utility facility for the movie Madagascar, the animators were able to render 50,000 hairs on one of the character’s heads (a lion), each designed to move independently when the mane was shaken. “Animators are greedy,” says Manley. “Give them more powerful machines and they’ll individually render every blade of grass in a [cartoon] lawn”.

To manage that demand, HP established a series of spot and futures markets for its animation clients, allowing them to effectively plan their consumption while also giving them the flexibility to increase purchases or sell off over-capacity as necessary. “It is this type of flexibility that you just cannot get through owning everything,” he says.

Such examples may seem quirky to most businesses, but have acted as the proof of concept for more core applications, says Manley. “To be taken seriously we need to take [the model] into mainstream enterprise computing,” he says. So for the last two years, Manley and his colleagues have been looking at how the system would work with software from enterprise application titan SAP.

Conceptually, the SAP enterprise resource planning suite is modelled through a series of six iterative transformations: from generic processes – such as customer relationship management for custom banking modules – to a full utility service, capable of being deployed to users and covering the full lifecycle of the deployment.

The final stage of that development is still many years off, admits Manley. But the principles of model-based automation, virtualisation and policy-based adaptation that are necessary achieve it are already in gestation.

And the business benefits of being able to forget software provisioning, licensing models and upgrades, in exchange for a service that will be comprehensive and infinitely flexible to meet requirements – and where the costs are directly related to demand – make the model a compelling proposition.

Adaptable software may seem like an answer for tomorrow, he says. But today, there are hundreds of computer-intensive areas where demand fluctuates, where it could be applied: Liquidity modelling in financial services, gene sequencing in biosciences, drug design in pharmaceuticals, he suggests. “Once businesses are comfortable with these, then it will really take hold,” he concludes.