When it comes to infrastructure issues, Dell is not one of those vendor names that leaps automatically to mind, except as a very competitive supplier of `additional bits’. You want to expand the datacentre with some more servers and the company is always there on the RFT list. But when it comes to systems management choices it is not the most obvious name that springs into the frontal lobe. 

But there is the possibility that this will change now, following on from the recent news that the company has signed an OEM deal with Egenera for its Processing Area Network (PAN) management suite. The system has been getting some positive things said about it, and Fujitsu-Siemens has already signed up to OEM PAN into
Europe.  

It certainly works in a way that maps onto where some think the world is (however slowly) moving. That is to break away from the traditional concepts of the Von Neumann architecture where a server is a processor plus memory plus disk storage plus operating system plus application, and an infrastructure has lots of those building blocks. Instead PAN goes for far greater granularity, with everything being considered an asset that is available to be assigned to tasks as the management software determines. 

That shows strong conceptual similarities to moves in other areas, not least of which are the growing number of SaaS offerings, IBM’s open experiments with `cloud’ computing and even the potential that is starting to be waved around by the likes of Microsoft and Google. In essence, it becomes possible to build Demand-Response systems, ones that react to and manage the tasks required as and when the demand is made, even if it is unscheduled. 

This is, therefore, one of a group of early steps along the road to utility computing and aggregated service provision. Which is all very interesting except that there has to be a modicum of doubt about Dell’s place in this schema. The company’s history, indeed its significant success, has been built on high volume sales in a price competitive marketplace. Past interest in any area of systems management has tended to be provided by contracted third parties rather than the company itself. 

The Egnera website suggests that Dell will be integrating PAN with its servers, but there has to be some doubt about whether an `out of the box experience’ will be possible or whether third party expertise will be needed to get the best out of the combination. It might well work fine for small installations, but in a larger infrastructure it is still normally true that a good deal of expert hand-holding, tweaking and tuning is needed at the very least.  

Neither Egenera nor Dell has that sort of track record, yet at least.

The enterprise desktop is most definitely becoming an integral part of the datacentre, a fact that must be true if only because of the number of companies now offering different types of desktop solution. There are virtualised desktops, PC-blade desktops and their associated thin clients, and there are the tools that allow the IT department to build, and therefore control and manage, the workspaces that individual users are provided. One of the latest to move in on this latter area is Dutch-based Getronics, with what it calls its Future-Ready Workspace. 

The fundamental idea here, as with all such approaches, is that key one of trying to balance at least the appearance of end-user `freedom of action and choice’ within an infrastructure that lives largely by being well-controlled and managed. The Getronics approach is to provide a reasonable level of end user choice – and let’s face it, in practice many end users do not require too much of that to be effective for the business – by building on standards.  

This is being built upon an existing `pick and mix’ communications and networking infrastructure from the product and service roster of Dutch telecoms service provider, KPN, which acquired Getronics during the latter part of last year. The upshot is that the KPN telecoms infrastructure components have been pulled together and placed under the Getronics brand, where they have now been integrated with the workspace tools and existing infrastructure components.  

The company then works with other partners such as Accenture, which provides the integration piece and Atos Origin, which is responsible for the datacentre, in order to build an end-to-end offering.  

To make this work, of course, requires building on standards and as this is all about the desktop workspace, Getronics has opted for the de facto standard of Microsoft for the desktop environment. This is then used as the basis of a three-tier development model with which end user workspaces can be built.  

At the base level there are a range of core services that every user will need and are therefore part of the underpinning workspace infrastructure. These include all the obvious candidates such as print, mail, standard Office applications, storage (including backup and restore), communications and collaboration infrastructures and a baseline security offering. 

Above that in the model comes a range of services modules from which users can pick and mix the functionality that will be required. Getronics is hoping that most users will be able to find modules for most of the workspace requirements they have, and it does create an obvious opportunity for the company to grow that module repository as user demand/popularity changes. The current modules mainly include extensions to the core services, particularly in areas such as communications, collaboration and security. 

The final level of the model is where user enterprises can develop their own specific modules, if required. This is aimed at those users that have specific requirements in areas such as security and collaboration. 

This end-to-end package is primarily aimed at the larger enterprises and, with Getronics being Dutch, Shell is an unsurprisingly obvious typical customer. But it is now looking at a SaaS offering which, while intended as an alternative delivery system to onsite installation, can also open up the workspace offering not only to small and medium sized business (SMB) community, but also the channel systems integrators and applications vendors that service much of this large market sector.

The Cold War left behind, if nothing else, some hot property. This is especially so if you are in the market for the ultimate in secure datacentres and don’t mind operating out of an isolated spot high on the Lincolnshire Wolds, on
England’s windy East Coast. That is what Lincoln-based managed services specialist, CentriNet, found when the company discovered an ex-Ministry of Defence Radar station - nuclear bomb-proof with 4-metre thick walls, some 30 feet under the ground, and with the bulk of a necessary refurbishment already done – was going for a snip. And what has this to do with anything? Well, in part it has to do with the fact, if any business is looking for a supplier of a highly secure hosting service – actually finding the place is hard enough, let alone getting through the security procedures – that is exactly what SmartBunker is designed to provide. 

But in part it also represents a potentially important little marker along the road towards a major seachange that I feel all IT users will need to face up to and learn to live with – Utility Computing. In addition, I suspect it can provide a useful testing ground for companies looking at how to integrate Utility services because it offers such a precise specialism.  What CentriNet is offering is a service that is well beyond the capabilities of most companies – finding a nuclear bomb-proof bunker and then equipping it properly as a datacentre is a budget-buster for all but the biggest players. But the other side of that coin is a simple question: if the investment needed was much less of an issue, how many business managers would feel the need to have something like SmartBunker? How many businesses would find it very useful – or at least comforting – if they could have their important business processes and data tucked into a corner of such a place? 

That capability – the chance for most companies to have available, on tap, a level of service provision and resources that is well beyond their individual capital investment plans – is the choice that Utility Computing will offer to business managers. In addition, they will be able to pick and choose from a whole roster of available services, balancing their business needs against their budgets. The downside to it all is the perception that, because they are sharing resources, they somehow will not be `owning their own data’. Well, logically, the more data that is kept in digital form the less companies `own’ it anyway. I suspect that the `owned’ data that is stored on most company infrastructures is far less secure than in SmartBunker today, and in any service provider’s facilities in the future – the security of customers’ data and operating environments has to be a core business process for them, otherwise they will die quickly. The same cannot be said about many companies’ respect for their own data. 

Couple this with the growth in managed services – CentriNet’s core business in fact – and you have the makings of a valuable test exercise for many businesses. At base level it is already a good testing ground for integrating a high security hosted environment. It is also a good test of off-loading the management overhead onto a managed, hosted service.   But that high security piece adds one more dimension that also offers a chance to test a switch to a service provision-oriented environment. Here the fact that this service is going to be rare, and beyond the means of most businesses, makes the ability to share its resources attractive. It will also make it, perhaps in the long run, an ideal platform for testing the service levels, capabilities and the operational integration issues, of running with a service provided as a utility.

The change in the nature of what we call `infrastructure’ is now well and truly underway, and while many large enterprises still feel it will not affect them directly, I would beg to differ. 

The as yet unofficial news that Microsoft is to invest in building some 24 datacentres, each of at least 500,000 sq ft floor area and costing around $500m a time, raises the stakes on even the largest enterprises having to seriously look at what this type of capability can offer them.  

There are, I suspect, still significant flaws in not only the thought processes of enterprise IT managements but also Microsoft itself in the development of its datacentre strategy, but there is no doubting the potential for change in the way ICT services are delivered to end users that Microsoft and Google are about to bring. 

Microsoft’s biggest issue, I suspect, is that it persists in seeing the SaaS operational model as a way of delivering applications to end users, rather than services. To me, that is still operating with the mindset that pens are about `the mechanics of ink delivery’ rather than `writing’ or going to movies is about observing `shutter/celluloid frame image integration’ rather than being `entertained’. This could explain one of the reported hindrances to the planned Microsoft takeover of Yahoo, where there seem to be issues about integrating, or not integrating, the two businesses.  

While there may be scope to integrate technologies at some time, as well as some of the operational concepts, integrating the brands would be silly. Any final takeover would give the opportunity to move all consumer-directed services (ie `applications’ in Microsoft parlance) to the Yahoo brand, while directing the Microsoft brand even more directly at  the business and enterprise sectors. 

It is the greater flexibility of the operational concepts possible with this datacentre growth that enterprise IT managements need to be aware and ready to work with. They may be thinking it won’t affect them as it is all `consumer stuff’. But some of that, the online use of Microsoft personal productivity applications that are the stuff of working life for most enterprise coalface workers, is likely to become endemic amongst them. It will be difficult for IT managers to stop it, I suspect, especially if those workers find it is faster and more flexible to work with an online remote service than with the in-house service. 

And where the individual application/service goes, so more complex services will start to follow. IT managements that don’t at least consider embracing SaaS delivery as an alternative to some elements of the infrastructure – if not the entire thing just yet – may well find it happening within their domain without their control or influence. As we have seen many times before, disruptive technologies like this always create situations where the line between `flexible adaptivity’ and `anarchy’ gets very thin and diaphanous. Those enterprises that can get into the former camp tend to win. 

One final thought: Microsoft already has BT as a favourite cuddling partner in the SaaS marketplace. Indeed, because of its background BT probably has a better understanding of the concept of `service delivery’ than Microsoft and its service=applications mindset. So it could be a good bet as a brand capable of pushing Microsoft into the business and, eventually, enterprise infrastructure space.  

It was not so long ago that Microsoft had some serious aspirations towards acquiring BT. Could it be beyond the realms of impossibility……….?