The increasing pace of business change has put network management into the front line. Only by understanding what is going on right across your organisation can you plan for the future. And only by linking IT to the business processes can you adapt the former, to best suit the latter.

The best known model here is HP's Adaptive Enterprise, which is a map for the development of IT systems which will automatically evolve and change, as the business processes that they support evolve and change.

While HP would argue that its software and services are the best choice for building an adaptive enterprise, the concepts are generally valid and there is nothing to say that they have to be HP-based. Indeed, IBM is among those with some similar products and concepts.

HP suggests the following four design principles for an adaptive enterprise:

Simplify: there should be fewer elements, with no customisation. Standards: use common architectures, and standard processes and technologies. Modular: develop logical architectures and reusable components. Integrate: build a managed and dynamic link between business and IT, connect up internal and external applications and processes, and automate change.

The final point is particularly important, according to Nick van der Zweep, HP's director of utility computing. "Change has to be automatic, it can't be rigid," he says. "You can't buy an adaptive enterprise, you have to build one so we give a reference architecture."

Whatever technology is used, network management software is an integral part of the adaptive enterprise model, because it fills the role of the operating system.

"OpenView is key to adaptive management," says van der Zweep. "The software layers we are creating are analogous to a PC operating system. You have a set of heterogenous CPUs, you have heterogenous storage, you have a network and applications, and the OS balances the different calls on the resources."

At the infrastructure level, the necessary standards and modular resources are enabled by virtualising the hardware, and indeed the entire data centre. This means bringing in grid computing, for example, and storage area networks (SANs) rather than direct-attached storage.

"We are betting that there will be multiple CPU types at customer sites for the foreseeable future, and that no one operating system will win out," van der Zweep adds.

"To manage large numbers you have to virtualise, you can't be bound by the hardware. It's virtual storage, virtual servers and virtual networks. Virtual software is coming - we are grid-enabling all our software, whether it's for mobile devices or the high-end Superdome."

HP calls this approach the über-OS, or super-operating system. Based on OpenView, it assumes the creation of an adaptive management layer that is an essential enabler for, and subset of, the adaptive enterprise, and which forms the link between business and IT.

It says that there are three main steps to adaptive management:

1. Manage assets and resources; 2. Automate processes, establish workflows, and link IT to business to communicate, measure and deliver services; 3. Prioritise IT actions, establish a business-oriented service management culture, optimise utilisation, and virtualise data centres.

To put this another way, business process monitoring feeds in at the top level, below that is application management and measurement, and under that is the infrastructure.

For example, a key infrastructure component of this über-OS would be software which allows administrators to

Just how autonomous (or autonomic) an adaptive enterprise could ever be is debateable, as human skills and judgement will probably always be needed. But it is undoubtedly a powerful vision of where integrated network management, virtualisation and grid computing could take IT in the future.