How many servers do you own? Where are they? What operating systems are they running? Windows? Unix? Linux? A mixture of both? What chip architectures are they based on? X86? Sun Sparc? Risc?

Over the last ten years distributed computing has brought IT much closer to the user, with the trend towards departmental application servers devolving power further in their direction, and driving IT to become more responsive to the needs of the business.

However, there has been a price to pay for this in management and maintenance terms. Because as server infrastructures have grown, so too has the complexity, time, effort, and cost of maintaining them.

As a result, all the questions above – and more – are becoming increasingly business critical.

This, in a nutshell, is the perfect sales pitch for server virtualisation.

Essentially a software layer that lets you treat one physical server as several different machines, server virtualisation is basically about reducing the number of servers the business operates; partitioning the computing power of those that remain to deliver the necessary level of service to each application.

At the top level then, like its cousins, server virtualisation amounts to consolidation; this time reducing the number of datacentres or distributed servers the business has to manage, and saving time and money in the process. But there are other benefits too. Improved system and application performance; faster application deployment; cheaper, easier software licensing management through simplified tracking… the list goes on.

Virtualisation also plays well to the “green” agenda.

Fewer servers means more efficiency, less wastage, and that less power is needed to run and cool them.

It also allows the management of different operating systems and applications from a single point, so resources and processing power can be moved around and focused where they’re needed most. And because it’s all virtual, there’s no longer any need for your people to physically load and test new applications on individual servers.

Perhaps virtualisation’s greatest benefit however, is that it allows for the maximum utilisation of each physical server and, in light of today’s usage, this is vital.

Analysts agree that, at present, few servers are utilised to much above 15% of their available processor and hard drive capacities. With virtualisation, it’s claimed that this rises to 70-75%.

Unsurprisingly given such a context, competition and interest is really hotting up in the sector. For now though it is dominated by the classic 800-pound market gorilla. VMware. Its products sell at a premium over its rivals, but many users seem happy to pay this in return for its perceived maturity.




This has seen VMware become almost the de facto standard in server virtualisation over the last few years, with the company having recently floated in the US with a market capitalisation of $1.1 billion.

Other big names are playing catch up as fast, but there are still gaps in their offerings. Microsoft, for example, has yet to announce a launch date for its Viridian virtualisation suite (and has already put back the date for the beta version) though it is expected to arrive sometime in the second half of 2008.

Elsewhere, Citrix recently announced the purchase of open source virtualisation supplier Xensource for $500m.

So how does server virtualisation work in practice?

If you’ve built up a distributed server infrastructure, you probably have file and print servers for different departments, email servers, web servers, database servers as well as all of the usual business application servers; all backed up with varying levels of redundancy. Different departments or functions will likely have varying resources and power depending on their function.

(You may even be consolidating already; migrating to a blade server infrastructure perhaps, or moving from Unix and or proprietary operating systems to Intel or AMD-driven Windows servers.)

With a virtualised server infrastructure, instead of a physical server being assigned for each application or department, the servers are ‘split’, enabling them to run several applications simultaneously. From an operating system perspective, rather than using multiple physical servers to run different operating systems, you use the same machine split into a number of virtual servers – each operating as a “single” entity. This can be scaled to complete server infrastructures.

Opt for such an environment, and the first challenge is planning.
It is unlikely that you’ll be able to migrate all your applications at once, so choose carefully which you move, and when.



The finance department is likely to be busy at month- and quarter-end, for example, so you won’t be thanked for adding any disruptions or further pressure. In short, timing is crucial.

Adoption will probably present some cultural challenges too. Users accustomed to controlling their own IT assets will need convincing that they’re not going to suffer drops in performance or resilience, for instance.

Many organisations opt for regression software here. You identify the application to be moved, evaluate how it’s performing in the virtualised environment including any performance issues, and then reload the application onto the original physical servers while any problems are addressed.

All in all, where updating your server set up is concerned, there’s really no reason NOT to consider virtualisation these days. It’s powerful, it’s cost effective, and it’s maturing fast.

Also, unlike many technologies, it hasn’t just trickled down from the high-end datacentre environment and appeared by default. It’s emerged fully-formed from the departmental and desktop server space.

And there’s nothing virtual about that.