Banks Statement

Get ready for a 'killer app'

The one trouble with most of the vendors of SOA solutions is that they are actually vendors of technology. As a result they usually fail to see the wood of end users' real needs for the plethora of technological trees with which they are so familiar.

In short, they usually sink back into talking about what is safe home territory to them. The trouble is, it is often inexplicable gobbledegook to everyone else that really should be listening and understanding the concepts for which the technology was created. One of the troubles with SOA technology is that it has become fearsomely complex to all but the four people in the world who really do need to understand it. To real users, however, people with business processes to run, it is meaningless drivel said in a foreign tongue. What they want is information that gives them some idea of what SOA can really do for them and their business.

In old-fashioned IT terms it needs the appearance of a 'killer application'. More specifically, it needs the appearance of a 'killer process' - a capability which gives users that sudden but blindingly obvious insight into why they do need SOA after all; in just the same way that the Lotus 123 spreadsheet - and its progenitor, Visicalc - made business users realise why they really did need the PC after all. And, as with 123, any `killer process’ will not find favour because it is `cool technology’ (and as an aside, I fear I might ritually disembowel the next marketing suit I hear saying those words). Instead, it will find favour because it provides business users with some functionality, some capability, that they did not have before – or least not in such a convenient, appropriate form – and that gives them little option but to admit that they need it.

Lotus 123 was such a good example of this: before it cam along users could work out the formulae needed, they could write down the numbers, but then they had to sit and do the maths the hard way. With 123 they not only got the results (OK, usually accurately) at the press of a RETURN key, but also could repeat it quickly with variations on the number set – the all-important and incredibly useful `what if’ game was born.

So what might that killer app be for SOA? Well, I accept it is still early days but I do believe I have seen a real contender in action. This is a research project dubbed `Secret Island’ being run at IBM’s Hursley Labs. It is currently based on Linden Labs’ Second Life online participatory video gaming environment. There is more detail in a piece I wrote for The Register (www.regdeveloper.co.uk/2006/09/21/ibm_secret/) but suffice it to say here that the use of the game’s avatar representations of individuals starts to make sense when applied to a business environment. I’ll admit I know little of Second Life itself, but I am old enough for that amount to leave me with the question – `why bother? Why not get a real life with real people?’

But to see those avatars representing individuals that could, for example, be in a project team scattered around the world, meeting in whatever combinations of participant that was required, gave it all a direct and obvious purpose. Add in the potential to link those individuals to the applications and services that they use, both within the business and without, and suddenly the `point’ of it jumps out of the screen at you.

To date IBM’s rocket-scientists (led by one Ian Hughes, who has a job title many geeks would die for - `metaverse evangelist’) have only proved the ability to integrate external services with one example – a link to Amazon. But one example does equal a significant potential (API development willing) and the whole point is that, while there is some serious technology going on underneath the current experiment and some even more complex technology required to create an SOA-enabled working environment that businesses could actually exploit, none of that seems relevant. I sat and watched the early demos and thought – `I can see how this could be used….. I could make use of this, even in my job’.

There are bound to be more candidates for SOA killer app. Indeed, this is likely to be one of the next big splurges for off-the-wall development efforts. IBM’s Secret Island may not prove to be the real killer, but it sure as hell shows that the time is right for one to appear.

My New Sub-Editor

Sub-editors are essential. All journalists will, under torture naturally, admit that this is the case. So let me introduce you to my new sub - TypePad.

For some reason I have yet to work out and, in the end, can't be bothered to spend any more time than I have already trying to, TypePad decided that two place names in my last posting - `Bourneville' and `California' -  had to appear on separate lines.

This is despite the fact that no carriage returns were inserted, ordered, left over from the previous post or any other of a million excuses that might be offered. I have tried everything I can think of to make those words play their humble role as integral parts of sentences but they insist on standing alone - out and proud protesting the right of place names to appear emblazoned and individual and not forced to mix with the common hoi poloi of ordinary words.

So I apologise for my sub-editor. If I could fire the recalcitrant little toe-rag I would but for now I appear to be stuck with him/her/it.

There is, probably, a reason for it happening. No doubt it is to do with my using some bit of Microsoft wotsit in a Java thingie without first coding the Lord's Prayer in PL/1. But after all these years of laughable cock-ups you would by now think that those that provide this stuff would be able to get a simple cut and paste from a word processor to a Blog entry to actually work.

I, and a million companies that increasing depend on this stuff to survive, obviously expect far too much.

The return of the `company village’

Anyone with a bit of interest in English industrial history will know of the Cadbury family and how it effectively established the town of

Bourneville

as a place for their workforce to live. That is a story of high levels of philanthropy and good intentions – there were other examples at the time where the end results were less attractive. The song `16 Tons’, with its chorus of  `I owe my soul to the company store’ was often not just true but painfully commonplace.

Could we be going there again, I wonder? The reason the question emerges has a number of facets, not least being the speculation surfacing at IBM’s Almaden Research Labs in

California

that there could well be a reversal of the long-standing trend towards urbanisation. I was out at Almaden a few months ago and this possibility was mentioned to me by Dr Ray Strong, the science lead for IBM’s new On Demand Innovation Services (ODIS) project. This is a service pitched at big clients that want to try and get some sort of handle – indeed any handle – on future trends and their likely impact. This is long range stuff as well – 20 to 30 years out – where futurologists usually meet SciFi writers in a game of outlandishing each other for wild guesses.

For what it is worth, Strong reckons ODIS offers a way not of predicting the future, but at least defining future sign post events that will point to possible futures. One such future being speculated is the said reversal of urbanisation. One obvious `sign post event’ is the development of the Internet and all that has followed. It is now possible for a growing number of people to do productive work totally remote from their `place’ of work. Location and value no longer demand proximity.

So they can live anywhere, but what about those people who still need proximity to a fixed location to do their job? Well, lots of us like the idea of un-urbaning ourselves – I have done it myself with much gratification. But not everyone can afford such luxuries, and as more aim at it, so the costs will go up, so pure economics might well keep an increasingly frustrated and disillusioned majority urbanised against their better judgement.

Then there is that other factor – the `Big House’ syndrome. One of the first companies in the IT world to adopt this approach in Egland was SAS Institute, which bought a huge and ancient `pile’ at Medmenham, outside Marlow. Others have followed suit and many of our grand houses are now occupied by Mammon’s labourers. In a way, nothing much has changed there: serving Mammon was the reason most of the houses got built in the first place.

And the grand families occupying those big houses often grew villages around them so that estate workers were on-site, so to speak. Many still own those villages lock, stock and undertaker. So why shouldn’t the big companies ponder the same possibilities again? They are the ones with sufficient resources to buy the land required and make the level of investment demanded. They and their staff also have a common need. The company village could be reborn.

Lastly the technology is already largely available, and it would make particular sense to utilise the capabilities that utility computing architectures could provide. There is a synergy there between technology, social need and brute economics that could make such an idea very appealing.

Then again, of course, it could be absolutely awful. Much would depend upon the attitude of the boss and the board of directors – and we can all think of a few we’d rather wasn’t our `landlord’, I suspect.

Is there a join between Open Source and Pay-Per-Use?

It is reasonable to assume that some form of Pay-Per-Use (PPU) is the model that will be adopted both by software vendors and by business users as the most equitable way of reimbursing the vendors for the value the business users get from their wares. It is also reasonable to assume that some form of Open Source licence model will come to prevail amongst the software vendors as a core component of such a model.

If one takes what Simon Phipps, chief open source officer at Sun Microsystems, was saying only yesterday and extrapolate it just a bit, the connection becomes quite clear. Phipps was presenting the formal announcement of Sun’s decision to move Java SE to an open source licence model, starting at the end of this year. This, he said, would be moving the point of monetisation to the point of value for the user – in other words, they pay for what they use, when they decide to start using it.

That is still some way from a micro-payment PPU model, but it is also a hell of a long way from the industry’s tradition of up-front payments for everything that can be thrown into the `package’ regardless of whether you want it or not. (Hands up all those that: a) have Microsoft Office, and: b) have never used at least one of the applications purchased, and have no plans to in future).

It won’t be a simple or direct relationship, of course. Open source is, for example, potentially more fair on the copyright holders of the code, which does not necessarily mean just companies. Individuals can be licenced code committers to a specific application code base. They are the people licenced to add functions or components to that code. Indeed, many individuals within major Open Source vendors are licenced in that way and, depending on the terms of the licence, may have a variable degree of conscious or even unconscious control over how an application or service develops. It is not unknown, I believe, for committers to have their licence to commit code revoked if they leave the employment of the company producing the product, but still retain their copyright over their contributions. That obviously means the potential for a legal mare’s nest is quite high. According to Phipps, trying to sort through just such a legal quagmire lay behind the time it has taken to get an Open Source implementation of Java SE.

So there are obviously issues to be sorted out with Open Source licencing in order to make a model that fits with PPU. But as that is the way that more businesses are heading (albeit slowly) and the way that even the most proprietarially-oriented software vendors are looking (witness Microsoft and its `talking up’ of Microsoft Live) serious changes will be demanded of current proprietary licences – changes that will inevitably move in a quasi-Open Source direction of charging at the point of `value’ (definition of which has the potential to be a legal quagmire in its own right, of course).

PPU is the payment model business managers are going to want, because it makes economic and account management sense. It will, therefore, push proprietary vendors towards licence models that increasingly mimic Open Source licences. In time, I suspect, the two sides will be so close that there will be no practical difference. Then, of course, they will hate each other even more than they do now……plus ca change…….

Terrorism as a market kicker for Utility Computing

I sit here in my little office, looking out over a garden raggled and taggled by a blustery day and think: thank whatever supernatural power you want to name for the fact that I am not trying to fly anywhere today.

Because of actual/possible/supposed terrorist activity (delete as the story unfolds) trying to fly anywhere out of the UK today is a hopeless prospect, and I feel for the poor souls trapped at airports around the world wondering what they are going to do now. And because I have done my fair share of flights lugging a laptop around with me, grateful for the capabilities it offers `on the road’ but cursing the weight and inconvenience, I have some feelings for any business traveller now caught in the dilemma of risking the valuable piece of kit in a suitcase jammed in the hold, or leaving it behind in a pile of banned detritous not allowed on board – at least until minds are changed about allowing hand luggage back on board.

But then I suddenly thought – perhaps there is some good that can come out of this in the long term. I cannot be alone in cursing all those people, busy `execs’ usually, who insist on dragging half a wardrobe on board with them as `hand-luggage’– usually plus a laptop in a bag. They think it saves time, but the delays they cause getting on and off usually use up any advantages. Perhaps, however, there is scope for making all aircraft hand-luggage-free.

Ahh, but I hear voices stage left muttering: `what about the essential laptops?’ Good question, and there is an answer. OK, it is an answer that is not quite deliverable just yet but: could the rise of the terrorist threat be a market kicker for the development and growth of Utility Computing?

Think about it: UC has the potential to create an environment where everyone’s IT `existence’ is held on the utility, not on individual PCs. All the user will do is log on and, verily, their work (or whatever) environment will pop up in front of them. Indeed, they could have several identities – for work, home, other nefarious pleasures – each of which would be equally accessible.

And accessible from what? Well, certainly not a laptop – too passé, too heavy, too bulky and too damned slow to be anything but an historical curio. For every day use I suspect there could easily be some form of lightweight, totally anodyne and technologically agnostic thin client system. But if people are in situations where even small, thin clients are banned – and we must expect that hand luggage will become a long-term no-no in aircraft, after today – then even they would be banished to the detritous pile. The only answer would be a rented client device at the point of arrival. And if everything about you and your work, life or whatever is on the utility system, not on the client, and only ever passes through the client without stopping in order to get to you, would you actually need to have a client of your very own?

I understand the psychological sense of security in having `your own’, but it would be a total irrelevance. It would also make air travel so much better and more convenient. And if you insisted on working on the flight then, “why sir, there is this thin client that pops out of the arm-rest.”

If the terrorists are achieving anything it is to point out that those of us who lug laptops around the world thinking we’re being very clever and productive are probably being the exact opposite. After all, how often have you been heard to yell, when in some far-flung hotel on the dark side of the moon, “why the *!!!*$$ won’t this laptop connect?!”

In practice, all we need is access to our work or environment. At the moment, perhaps, it means we need to carry that around with us, but we are within a whisker of never having to need to again.

DIY technology forecasting

If you promise not to laugh I’ll tell you a little secret: for many years I have been interested in the ancient art (if art it be) of the Runes as a guide to life, the universe and everything. Most people use them as a guide to their personal lives. Some people put far too much faith in them.

But it recently struck me that people in industry put similar amounts of faith in the outpourings of the market and technology analysts which, despite all their talk of models and methodologies are just modern rune-reading diviners – or should that be guessers? The only difference is that their runes are made of mangled numbers rather than bits of wood and bone.

So why not use an ancient divination `technology’ - the Runes – in a new context such as answering important questions about IT technologies, markets, models and trends. For example, how about one important question of the moment: is the way IBM and HP are buying up SOA-related technology as fast as they can going to be a benefit to users in the future?

At the moment the SOA business is the mixture of apparent certainty displayed by these companies about what should be happening and where it is all going, set against the confusion that seems to exist in the minds of both users and the wider community of vendors that are, or which claim to be, providers of `SOA solutions’.  This actually shows through in the two runes that turned up representing the recent past, and the present.

For the Recent Past there came Sowilo, the Sun Rune, which denotes power and strength. That which you want may be attained. It also denotes mental clarity and added warmth to your relationships.

Here is that recent past of IBM and HP in the SOA arena, giving the subject plenty of sunshine, talking it up to anyone who will listen, and with sufficient clarity in their own corporate minds that they do attract the interest of quite a few potential customers. At least the customers are not rejecting it out of hand.

For the Present came Isa, the Ice Rune. This represents stagnation and a passionless existence. In practice, it shows that your life's course may seem blurry at the moment, but if you persevere you will move onto better days.

Well, if that doesn’t sum up the current state of play in the SOA world I don’t know what does? Despite the talk there is not too much happening, with a fair amount of sitting on hands and fences by the customers. The way in which SOA will develop as an operation component of a business infrastructure is still not clear for many of them; though if they don’t turn their back and run away from it there should be valuable results to be had. In the case of IBM and HP, it is possible to see both as casting around in the blurry course of their lives, snapping up anything that might help give some clarity. Their collective need to persevere will show itself in the effort they will need to put into making the acquisitions gel into a working whole.

It has to be said that, normally speaking, both companies are pretty tenacious at the persevering business and an indicator of the consequences for both the vendors and their customers may well be seen in the rune for the near future.

This is Othala, which represents a solid, immovable home, prosperity and safety. Good fortune based on your heritage and character is yours to enjoy.

God in heaven, does that mean the old adage – no one ever got sacked for buying IBM – may still hold true? Well, let’s face it, for any user to move into an SOA-based infrastructure will be taking steps that are well beyond the scope of mere `technology’ – and smart-ass new technologies are all most of the other vendors are offering. This the same old dance in which the IT industry indulges: if in doubt, buy something new and shiney.

The technology is important, but it is a minor part compared to understanding business models, the integration of those models and the technology and range of other factors. The likes of IBM and HP are amongst the few companies – the solid, immoveable ones – that actually have the depth of understanding across both camps to build the necessary bridges. That is why they are buying in technology now – and I would expect to see them starting to discard the unwanted or superfluous bits over the next three or four years. Not all of what they have bought will be needed for those bridges, but to claim that they know now which are the relevant technologies to keep would be little short of guesswork.

As for the Distant Future, there is the Rune Gebo, which represents the coming of gifts and favours which can also carry obligations. This rune often represents strong bonds such as deep friendships and marriage.

Whichever vendor an enterprise chooses as its primary contractor for any move into serious SOA it will have to be for the long haul – this is not popping into the supermarket for a can of beans. These relationships will be like marriages – and no doubt with the same ups and downs. Any business relationship like that can bring seriously good gifts and favours, both for the vendor and customer, but it sure as hell will bring some serious and ongoing obligations. The SOA vendor will become increasingly embedded as a part of the business infrastructure of its customer, with a degree of mutual dependency that may go further than either party has experienced before. So the vendors that have the depth of technology on which to build solutions that meet every users’ needs, and operate within a culture that understands a wide range of different business practices and how to service them, is best placed to meet the needs of the marketplace.

There: some of those analyst companies would charge you $250,000 for that.

SOA `prime time’ needs pinches of salt and sanity

So 90 per cent of organisations are introducing or planning to introduce SOA according to a report by analysts, the Aberdeen Group. It is reported that the Group also suggests that making such moves will eat up to 40 per cent of the typical IT budget. Hmmmm, a pinch of salt might need to be taken with that, I suspect.

For a start, no matter how keen I am on the SOA as the way forward, I suspect that a good deal less than 90 per cent of organisations are as close to it as suggested. If the `planning’ bit also includes thinking about thinking about it, then I’d accept the number – I’d think about thinking about the earth being flat if reasonable evidence started appearing. The evidence that SOA is a good idea is already widespread. It is also the way most companies – and individual users - will go in the long term, but that does not mean that many companies are that close to implementing it.

And as for consuming 40 percent of the IT budget, well: as it is generally accepted that 75 per cent is already consumed by maintenance of existing applications and infrastructure that makes a mathematically improbable 115 per cent budget. Alternatively, if it is 40 per cent of the remaining 25 per cent – just 10 per cent of the total budget – that could be considered a damned fine investment for the benefits in increased flexibility and speed of response to new business challenges, which is where revenue growth is most likely to be found.

Ten per cent is also a reasonable rate of investment in SOA given the way implementation is likely to go. It is a warm day in

Antarctica

when I agree with Microsoft top people by I would agree with Ray Ozzie, the company’s putative visionary in chief, that SOA will only be an addition to existing infrastructures, not a replacement. Well, to be fair, I agree only up to a point. He is right for the short to medium term – say the next five to eight years – but after that he’s wrong.

Yes, SOA is going to start life as an additive to existing systems. It would be a very foolish company that set out on a whole-scale rip-and-replace just because SOA is a `good thing’. But that additive nature will diminish as existing infrastructures come to the end of their useful lifecycle, and I know Microsoft understands the likelihood that a reducing number of users will be buying outright licences in future years. That business model for software vendors is doomed.

What follows it is still the matter of conjecture, and there are many attempts at generating new models starting to appear. One such is DreamFactory, which describes itself as a `leading provider of adaptive on-demand applications’. (As an aside, I wish to God companies that are in a brand new, untried market would stop describing themselves `leading’ anything – except perhaps in marketing bullshit).

In practice, what the company is doing is providing a set of applications and tools via the salesforce.com AppExchange service. I am not going to venture an opinion on the products themselves – one must assume that they have been sufficiently tested to at least work together in any loosely coupled composite SOA environment, as well as work with any other application or tool in the AppExchange roster. Without casting any judgement whatsoever, I accept that this may be a major and possibly foolish assumption for any IT manager to make.

But setting that proviso aside, this model of offering components specifically designed to be elements in a composite service is the way most software vendors have to move. In addition, the AppExchange approach shows signs of being an embryonic service aggregator, the type of operation that will, in a future where utility computing is far more widespread and normal, be the primary source of services for most users.

There is much that still needs to happen before we get there, of course. Not least is the fact that AppExchange charges out on a monthly subscription model, which can often be just a deferred payment version of the buy-the-licence model. Not till payment-per-use, and indeed micropayment-per-microuse, comes along will outsourced services make a good business model. Till then, Ray Ozzie will probably stay `right’.

We all hate standards, but……

One just has to look at what is going on in the world of web services and the web generally to see that standards, hateful and creatively restricting though the little devils can be, are actually a good idea. They may cut short the life of some clever bits of technology that don’t fit in, but the results that can be achieved with the collection that do can be remarkable.

So, the fact that the Open SOA Group has got itself to the point where it is ready to disport itself publicly with a new website (www.osoa.org) and has a bunch of specifications available that are designed to make the creation of composite, loosely-coupled services a great deal easier, is something to crow about – up to a point at least.

For example, the specifications in question fall into two main groups: the Service Component Architecture (SCA) and Service Data Objects (SDO). SCA covers a range of modeling tools – how you loosely-couple applications in a consistent manner so that creating the composite services does not – or damned well should not – require too much in the way of hand-tooling to make them fit together.

SDO covers how data is handled between different applications. Again, consistency across a composite of different applications is the important goal, and this one is somewhat important – losing data on the way between apps would not only be embarrassing, it could be fatal to the business.

One of the most important points sustaining the OSOA Group idea is that, no matter badly the member companies might lust after total dominance of any user’s service provision infrastructure, the real world is different. Indeed, the most effective services any user creates will almost certainly incorporate applications from a number of different vendors. So having them all work to the same interfaces and protocols – the same standards – will inevitably be a significant advantage all round. Web services have already shown what can be achieved.

If such specifications are important – and I believe they will be, even if it is not exactly these as currently proposed – having all the major players onside would seem to be a sensible objective. The current membership list quite impressive, with BEA, IBM, IONA, Oracle, SAP AG, Sybase, Xcalia and Zend now joined by Cape Clear, Interface21, Primeton Technologies, Progress Software (formerly Sonic Software), Red Hat, Rogue Wave Software, Software AG, Sun Microsystems and TIBCO Software.

So it is a bit sad, though almost inevitable, that two of the leading names in SOA, HP and Microsoft, are not part of the mix. OK, Microsoft is not really a major name in SOA yet, but it has significant expectations of its place in that panoply, and HP is certainly a major contender already. It will be interesting to see how they jump and in which direction. If they don’t join in they will, I suspect, need a very good reason – and starting their own, or standing separately, will not be a good enough answer.

Microsoft using `right’ tactics on wrong point-of-pain

For a major corporate vendor, like Microsoft, `owning’ the customer will undoubtedly be considered the `right’ tactics. It has certainly done the company proud over the last 15-20 years for it now owns the vast majority of desktop and laptop PC users around the world. It therefore makes sense to try the same tactic with the enterprise market, particularly as it moves to a new, virtualised environment.

The trouble is, it does look as though the Redmond Ravishers are applying the tactic to the wrong point-of-pain enterprises will be feeling as they move towards virtualisation. The benefits of the technology are now pretty clearly understood, not just in the cost of ownership terms that niggle away at the souls of IT managers, but also in the increased flexibility and speed of response to changes in markets with which business people will most certainly identify. But most of the technical issues are pretty much sorted now, particularly in Microsoft’s backyard of x86-based servers, and the `sorting’ has been largely done by one company, VMWare.

With Microsoft some still some two years from coming up with anything that would constitute a viable contender for virtualising large x86-based datacentre installations and VMWare already several years down that road, it would look as though the Big M is fighting a totally-lost cause. By the time it gets there VMWare – especially now it has the clout of EMC behind it – will have moved the virtualisation goalposts a long, long way away.

The company’s response, as well analysed by my Register colleague, Ashlee Vance, has been to cut a deal with open source virtualization technology vendor, XenSource and suddenly decide that open source is a damned fine way to do business after years of claiming it was the spawn of the devil. This does look remarkably like a resuscitation of an old trick that was, for many years, the speciality of IBM. In essence, this meant that, every time a rival announced a new product or technology, Big Blue would announce that it was going to do it too, so as to spook users into holding off on the interest and orders, forever try to lever user emotions around that old saying, `no one ever got fired for ordering IBM’.

Given the fact that all is fair in love and market dynamics this is still a reasonable tactic to try, but Microsoft is, I feel, pitching at the wrong target by trying to muscle in on the virtualization piece with someone else’s technology. Important though it is, Virtualisation is just a component part of something much larger. It is also not the piece through which ownership of any enterprise marketplace will come. That piece is the infrastructure management system – and the higher up the management food chain it can be pitched, the tighter the user lock-in will be. Put simply, this is almost inevitable that the levels of complexity involved in such infrastructures will be too great for most user CIOs to do any more than say to a vendor: “Good God, you do it, where do I sign?”

This is, at least in part, why EMC snapped up VMWare, filling out it aspirations in that management arena. It is why companies like IBM and HP are so often the prime contractors on large scale, long term infrastructure implementation projects, and it is why a recent survey

showed that Dell is not a favourite server vendor amongst enterprise users despite its obvious skills at marketing and manufacturing management – not much of a track record at the management piece.

So far as it goes, of course, Microsoft hasn’t got much of one at the infrastructure level either, but it does have Systems Management Server (SMS) and Microsoft Operations Manager as building blocks on which to construct something. It also has the money to do exactly what IBM, HP and EMC have done – and continue to do - in building their own management offerings……..go buy the companies producing the right complementary technologies.

That is the area where it can come up with a management package that could give it ownership of a lot of users, for time is also on its side. Yes, IBM, HP, BMC and CA are already well-established in infrastructure management, but for now it is a market largely made up of only the biggest users. There is a vast army of second tier users yet to truly contemplate virtualization. But as soon as they do they will hit the wider issues of managing all those virtualized servers as a single information management and business processing entity; and entity in which virtualization is but a mere artisan in one corner. And they will try virtualization over the next couple of years as the power of multicore, off-the-shelf servers rises inexorably.

That is both a market that will be worth `owning’, and a market already psychologically geared to being `owned’ by Redmond’s finest sons and daughters. Virtualisation? Let VMWare have it.

How important is a Blade Architecture?

The short answer to that question is `very’. But the slightly longer answer is `only up to a point, Lord Copper’.

Following on from HP’s launch of its C-Class Blades, using a new architecture, Sun has joined in with the lunch of a new range of Opteron-based Blades, collectively known as the Blade 8000 modular system,  built to yet another architecture. I don’t plan to get into arguing the toss about which: Blade 8000, C-Class or IBM’s BladeCenter, is the best in architectural terms. This will always be an area for endless, unresolvable debates driven as much by prejudice – “IBM over my dead body/the only solution” (delete as applicable to you) – as they are by technical or economic efficacy.

I will just say, however, that I still think these Blade guys should get together and agree a real common architecture rather than preen and pose over which is `best’.

What is more to the point here is that Blades of whatever architectural orientation are little more than a clutch of diddy servers if there is not one other component in there as part of the mix. `Component’ is, of course, a rather demeaning description of what I have in mind, for it is the unavoidable need for reliable and comprehensive infrastructure management tools. Without this in place, an IT infrastructure made up of Blade servers is probably more of a mess than what it has been designed to replace. It might well be better for users to consider opting for a few of the new 16-way Galaxy data servers Sun has also announced.

The company’s track record here has been as the company playing catch-up to the likes of IBM, HP, BMC and CA. All of these have comprehensive – and growing – infrastructure management suites which go beyond the basic management and software provisioning of servers. This is largely where Sun lies with its N1 management tools.

The company is saying that N1 now includes comprehensive infrastructure lifecycle management, which can manage multiple systems across a datacentre, which is a move in the right direction. But in a world where the delivery of loosely-coupled, composite services running in virtualised datacentres built on several thousand discrete Blades, there may be room for doubt about its actual capabilities.

What is more, the very fact that N1 is now to be bundled with the Solaris 10 operating system, the PostgreSQL open source database, the Java Enterprise System and Secure Desktop in a package called the Solaris Enterprise System gives a pointer to where Sun sees the world – a very Sun-centric, Solaris-centric place. The real world of IT infrastructures is an inevitable melange of kit from all over – something old, something new, something borrowed (or at least open-sourced) and something (big) blue. Any virtualised datacentre environment is going to have to provide that range of service capability, and any management system worthy of the name must be able to manage anything the typical IT department will want to use.

If Sun wishes to be seen as a major player in the global IT infrastructure provision and management business these will be issues it must be seen, very publicly, to be addressing. For now it looks like it is painting itself into the corner of `we handle Sun servers best’, which will no doubt be music to the ears of Sun-centric IT shops. The rest of the world will, however, have moved on.

One last aspect of all this which could be of concern to IT managements is the fact that Sun has taken to the open source ethos with great enthusiasm. So the Enterprise System will be given away and company CEO, Jonathan Schwartz, has been quoted as saying that the challenge has been finding “the right mechanism to monetize the volume out there.”

When it comes to monetization the software may be free for a while, but the support will still cost. But his statement suggests that there is still some hidden, undetermined charge that could land on an IT manager’s desk at some (almost certainly) inconvenient moment. That may not seem like a good way to run a railway for many major enterprises.