So the best that HP can say about OpenVMS, as it passes its 30th anniversary as a front line operating environment, is that the software still doesn’t have an expiration date. This rates as one of the company’s greater examples of understatement, particularly for a system that seems to offer so much potential to the world of financial transactions in the `Web 2.0’ whirlpool.

OK, it is 30 years old and by definition totally decrepit, yes? Well not necessarily. It would be easy to assume that VMS-based systems reside in some technology backwater owned by companies with too little wit or too much conservatism to invest in new technology. But in practice many of these systems are in service – with HP not daring to call time on them – because they still fulfil an important, business critical role. And nothing has yet been found to supplant them.

That role is running real-time financial transaction management systems in some – quite probably most – of the largest financial institutions around the world. You can have your clever mashups of bits and pieces using Web 2.0 technologies, but when it comes to managing the real numbers of most peoples’ everyday life it is still 30 year-old technology that proves to be the best solution.

What is more, the capabilities of VMS – such as managing fractured, partially completed transactions through to completion – is just the type of financial environment that web-based commerce creates. Indeed, it is almost certainly what many of the mashups of Web 2.0 are most likely to produce if they ever front-up full e-commerce applications.

So rather than damn VMS with faint praise by telling the world it not yet signed the death warrant, why not be bold and push it into a marketplace where there is a definable need? And if VMS is too difficult for most IT departments to understand – it is certainly going to be a strange alien for the Microsoft-generation to deal with – then why not package it as a complete `black-box’ appliance? These days an Itanium-based server, be it a rackable brick or a blade system, with a complete implementation of VMS and enough capacity to run an application must be easily possible. And industry-standard connectivity should now make it a pluggable, interoperable option for many a growing e-commerce site starting to face the many issues of transaction management.

The cry is then often heard that appliances don’t sell – no one wants them. Well, for a lot of applications this is certainly true, but it is not a truism. Take, for example, the network router or switch. This is nothing more than a packaged, application-specific server appliance performing a dedicated, definable task that is both damned difficult for an IT department to engineer from scratch, and damned convenient to acquire off the shelf. That is the perfect definition of a suitable market for an appliance.

That is a market which, arguably, maps well on to transaction management – who cares what the front-end mashup systems are or the back-end databases if there is a proven, reliable transaction management box in the middle? But without such a system available off the shelf, IT departments are likely to find themselves increasingly obliged to try and engineer one from scratch.

A few weeks ago came the news that VMware, already well established as the poster child of enterprise server virtualisation, was extending its reach down into the small and medium-sized business world. There is a sub-text to this announcement, however, that opens up the world of Software as a Service (SaaS), not only to the small business user but also to their enterprise equivalent – the small department. Indeed, it has the potential to open up SaaS to major enterprises in the same way they grew to love Linux – by departmental stealth.

SaaS is already getting to be of interest to smaller businesses – as witnessed by the success of offerings like NetSuite, Salesforce.com and many others – but there is still a long way to go for it to have claimed to fulfil its obvious potential. One reason, I suspect, is because there are a large number of potential users with established, legacy applications wondering how they get from where they are to the brave new SaaS world.

`Legacy’ in this context actually means that panoply of applications and office productivity tools that run on Windows. These are not only the staple diet of the small business, either: they are also the bedrock on which most large enterprise departments are built, which means they have the same scope as small businesses to consider SaaS as future delivery option.

As one of the potential stumbling blocks for both parties is likely to be the perceived problem of how they get from the `here’ of on-site applications to the `there’ of hosted, managed SaaS delivery, the ability to re-use existing legacy applications could be just the bridging mechanism they require. With VMware claiming that just about anything that can be run on any x86 architecture can be considered as potentially viable in a small SaaS environment, then much of that fear of doing something new can be removed.

As one of its touchstone little tests, VMware often uses an ancient PC game from the Windows 3.x era, and this has been found to run as an effective deliverable in a multi-tenanted SaaS environment.

Yes, there are likely to be some issues here, not least with seriously non-trivial details such as licence rights on legacy applications in a new environment. But the ability to use VMware to create a small, multi-tenanted SaaS environment re-using existing applications has the potential to provide a very useful start point for not only the small business, but also the enterprise department; and it has often been the case that those departments have been the proving grounds on which significant new developments have been tested.

A couple of recent stories about Google have caught my eye, if only because they each carry an object lesson for users, and vendors, about the way the world is changing from the application of technology and to the delivery of services. What is more, together they show that companies like Google understand the trends and changes better than most of the vendors.

For example, Apache would be high on many lists of software choices for running a Web Server, yet its market share has started to drop away despite solid growth – 5 per cent a month - in new websites. Its share is down from over 60 per cent at the beginning of this year to under 48 per cent now which is odd, especially as many of the traditional ISPs would be amongst the most likely customers for Apache.

So where has the market growth gone for website services? It appears the answer is the social computing sites such as MySpace, Microsoft Live and Google Blogger. OK, it is fair to observe that there is likely to be a good deal of difference between the website needs and aspirations of a major global enterprise and an archetypal `spotty oik’. The former is unlikely to find too much in the way of suitable capability or resources available from these types of company, right now at least.

But the history of the IT business has already demonstrated that the technologies and functionality first developed for the personal/consumer/social marketplace – from high resolution graphics PCs onwards – usually end up somewhere in the knitting pattern that is enterprise infrastructure. The key thing with this trend is that the new market looks like it is coming from those that have little direct interest in the technology for its own sake – for its inherent `coolness’ (and I am trying to promise myself not to decapitate the next person who says “this technology is soooooooo cool”). Instead they have arrived at wanting a website through seeing others in action. Setting the sites up is increasingly easy, and nearly all the work is performed by the service provider, with as much of it automated as possible.

All the above companies provide social computing services rather than technology, which is an exact analogue of the way many new markets will develop in future. It is also why much of the growth in information-related services across the board is likely to come through users having the opportunity to interact with such services without serious upfront investment. If they can look and play – as per MySpace or Google Blogger – they will learn. If they can then build services to meet their needs both easily and quickly, there is strong chance they will.

It may start with the small business community, but many-a large enterprise department operates in exactly the same way, and the impact of such viral infections on the enterprise can be significant – witness Linux as just one example.

This is where Software as a Service (SaaS) comes into its own as a delivery vehicle for such a model, and let us not forget that Google is already the biggest operator in the SaaS business by a country mile, so it is not `the future’, it is  the `now’. And Google is doing all it can to make sure it happens.

To keep up with demand the company is well aware that it will need the latest, most powerful systems around, and for the future that means parallel processing. But that is only half the story, of course, for parallel hardware needs parallel software. So, in conjunction with IBM and budget of $20m-plus each, the pair is investing in providing a parallel clustered system to which a select group of US universities will have access.

The idea? Learn how to exploit the hell out of such machines, particularly when it comes to providing online services to a growing army of not just individual users but increasingly serious business users as well. That is likely to be the nature of `infrastructure’ for the majority of users in the future which means that, in practice, many of them will be able to forget about infrastructure altogether.

One of the fascinating things about rules is just what you can do with them once there is a convenient way of making them work automatically. That means rules engines and, while it may be easy to assume that automated rules processing and management is of necessity still quite simple, there is a subtlety growing in that business which is showing all the signs of moving beyond the ability to run just a process, and onto the ability to genuinely manage complex issues.

Take, as an example, some recently undertaken work by rules engine vendor, inRule, on behalf of an outsourcing and remote hosting service provider. It is easy to imagine where a rules engine would fit into that environment. It would be monitoring such areas as the capacity and predicted loading on available resources, for example, so that the system kept running as well as possible. It would be comparing contracted service levels against current and predicted levels given the overall upstream workload and capacity available.

The objective here would be, not unreasonably, the protection of the business of the outsource/hosting company – make sure the systems do not: a) dig a hole into which the business falls and, b) do not continue digging furiously if a hole ever gets started.

What this can mean, of course, is that when such a hole is spotted, the system will self-protect in the following order: itself; those priority customers with the most stringent Service Level Agreements and/or most punitive penalty clauses; the rest.

This is not a criticism, it is a fact of life and, in practice, it will often be the case that such a hole appears and grows at a rate which does not allow a company to do anything to civilised – like give `the rest’ a reasonable level of advance warning.

But what if it could? What if it was able to warn customers of an impending problem? It could even set up a situation where those customers could themselves take actions that reduce the possibility of such a hole getting deeper. This is the area inRule has addressed. In effect, the rules that manage the internal operations of the outsourcer feed results to additional rules engines that work out what status advice should then be given to customers. Such advice will obviously be driven by the resource utilisation demands and SLAs contracted with individual customers, but even for the lowliest of them the impact can be useful – in both directions.

For example if a lowly customer, with the worst SLA terms, still gets a warning that its service may go AWOL in three minutes it has time to make a decision – risk running an upcoming large process or defer it? OK, late running could upset its own customers but that would be better than dropping them down a hole not of their making. And the act of deferment could be just what is needed to reduce the demand on resources and keep everything operational.

It seems that there is considerable scope for infrastructure management to move to the next stage – the ability to manage automatically the ebbs and flows of workload through a growing complex of nested rules, each one feeding the next in the chain, both down it and up.

Somewhere, I suppose, such an environment will ultimately end up with a Tolkien-esque world where there is One Rule To Bind Them.