The race for leadership of parallel processing into the future is now well underway, and there is a chance that Sun Microsystems – not exactly a dark horse but not necessarily the first name to spring to mind as a main contender – may actually steal a march over the rest.  But this does then beg the question: how much will this directly affect enterprise users? 

It’s upcoming Rock Processor makes use of an old technology, transactional memory, but in a new way that should make developing applications and services a good deal easier than otherwise with any parallel software development technologies, such as multi-threading so far available. What is perhaps more important, however, is that Sun is also amongst a number of vendors funding, to the collective tune of $6m, Stanford University’s new Pervasive Parallelism Lab. The others include IBM, HP, AMD and Intel. 

It does not, however, include Microsoft, which has teamed with Intel to fund
Berkeley
University and the

University of
Illinois ($10m each) on similar parallel processing research. Microsoft and Intel are, of course, prime movers in multi-threading. 

Transactional memory can be implemented in software or hardware, but both can require significant levels of skill. Sun has come up with Hybrid Transaction Memory (HTM) which sets out to exploit the best capabilities of both, making development work easier than extremely difficult, and offering the chance for applications performance to improve without change as the Rock processor is developed further. 

And as the Stanford lab is headed by Kunle Olukotun, who helped design chips that are part of Sun’s UltraSPARCTx family, it might be possible to guess the general research direction the Lab may take. As IBM is also a chip maker in its own right, transactional memory could get some reasonable leverage behind it, which might just leave Microsoft ploughing a different furrow alone.

But back to that question: fundamentally, how much should an enterprise care about such things? Well, if one assumes that traditional processor and software architectures are, while by no means dead yet, widely diagnosed as terminal; and if one also assumes that what follows is parallel processing,  then every CIO and IT director needs to be considering what may be coming next, what its impact might be and what will be involved in implementing it. 

These are archetypal `big questions’ with no easy answers as yet. There might just be one, however.  

Parallel systems are going to be large – the larger the better in a way. They are also going to be expensive to buy and to run. But they are also going offer prodigious levels of performance, both in the shear grunt of throughput and in areas such as I/O management and capabilities.  

As we all know, the likes of Google, Amazon and Microsoft are building huge datacentres aimed at the nascent utility services market. They would likely find parallel systems strangely attractive, precisely for the above reasons. Then they would need to market the hell out of the service offerings.  

This could lead enterprises to seriously examine the possibility of switching to utility services simply because the cost and hassle of trying to make the parallel move alone will probably prove too expensive and/or traumatic. 

It could be that no-one need worry about making the move for it will only be the big utility players that have the half-dozen rocket scientists that ultimately need to know all this stuff.

The other week I wrote about the impact of the poor economic situation and the scope for really quite disruptive change in the building blocks of IT infrastructure that might follow as the world gets ready to spring out of the doldrums. It looks as though the IT vendors are gearing up for that eventuality, as new possibilities for inclusion in future infrastructure services are now appearing. 

Here are two of the latest that caught my eye. They are aimed, at least in part, at grabbing a quick-fix step-function jump in capacity and capabilities once the economy turns upwards.  

The first is another contribution to the latest fashion of datacentres in a container, targeting the need for finding quick ways of getting capacity without some of the hassles of installing and commissioning significant systems upgrades. 

This one comes from HP, which has christened it the Performance Optimised Datacentre – or POD. In much the same was as IBM and its container offering, HP is aiming to meet customer needs rather than provide an alternative sales option for its own systems. Users can specify any systems they like, so long as they fit in a 19 inch rack. With a planned turn-round time of six weeks between order and delivery, and a maximum capacity of around 3,500 compute nodes, this is one approach that may well have some short-term attraction for users looking for a quick scalability fix. 

The second comes from hosting company, Bytemark and, although primarily designed as hosted services engine, looks like it could also be the basis of something a little more permanent, even for in-house datacentres.  

Hosted services are not only an obvious option for meeting the requirements of rapid scalability, but also an increasingly viable option as a permanent alternative to many-an onsite datacentre. This is particularly so as server performance and virtualisation technologies continue to improve. But at the other extreme there is also the possibility of exploiting large numbers of standardised, simple servers, particularly in the Linux-running marketplace – and this exploitation can make sense both for remote hosting businesses and in-house datacentre users. This is, after all, one of the selling points of IBM’s z/10 mainframe, which can turn itself into a thousand or more virtual Linux servers. 

Bytemark’s approach is radically different to the z/10, but the potential is very similar. By using Intel’s Atom processor rather than the Xeon or AMD Athlon, it has been able to offer a low cost (£45 rental per month), reasonably-performing server, with 2 GByte of memory and two 100 GByte disc drives. Best of all, the thing only consumes 25W of power, which is a good deal less than most other systems worthy of the `server’ title. 

This type of low-cost, low-power, reasonably-performing server has the ability to hit a sweetspot for a large number of users in the Linux systems marketplace. As well as IBM, this market has been noted by HP, where very large volumes of low-cost Linux servers can be managed by a single NonStop server. This maintains a high availability server state record for each Linux server – effectively a record of what task each is performing and what applications or tools are running. Should one of the servers fail its workload can be shifted to another and the failed system replaced with no down time and a minimum effective systems redundancy requirement. It has the potential to be cheap to run, standardised so cheap to buy, easy to manage and very scalable. All of these attributes, and the latter in particular, are going to be important capabilities when the economy picks up again.

Musing on the state of the economy over the weekend I recalled the comments of AT&T cloud evangelist Joe Weinman who suggested at the recent Structure 08 conference in San Francisco that, when it came to the future of utility computing and `The Cloud’, “if the software licensing model doesn’t catch up, people will either be forced to pull from the open source world or the whole idea is going to drop dead under its own weight.”

Well, up to a point Lord Copper. I can see where he is coming from but there seems a sense of doom about the opinion that I don’t see and history, together with the cycles of economic fortune and more specifically misfortune, may even bear me out. For example, I got to thinking about the current economic slowdown and remembering others in the past. 

Economic times were pretty bleak in the late 70s when the first big disruptive technology, the personal computer, first started to appear in volume. Most of the mainstream IT vendors said it would unlikely claim a significant place in business. But who is in business now? Then the doom-laden days of the late 80s, epitomised by Black Wednesday, brought Standard High Volume (aka PC based) servers from the likes of Compaq and Dell right into the heart of enterprise client/server infrastructures. 

Now we have another economic slowdown and it is certain to impact growth in some sectors of the enterprise IT marketplace, and will provide an added spin to the fortunes of others.  

The standard response for most enterprises is likely to be a push for greater savings, so any short term investments are likely to be geared to `green’ issues, with a concentration on energy consumption savings rather than the more altruistic notions of reducing carbon footprints.  

There is also likely to be some push towards anything that gives users more Bangs for the same number of Bucks. So virtualisation is certain to grow in popularity. That is likely to bring a need for greater investment in new server platforms. These will not only be more suited to running virtualisation code, but also provide a jump in performance, a reduction in energy consumption and a reduction in maintenance costs that, collectively, should make the new investment worthwhile. 

But all that could indeed run up against the strictures Joe Weinman outlined. Yes, if the mainstream leaders of the ISV community don’t change their licencing policies, and change them soon, it could easily bring the potential of The Cloud to a grinding halt. But it could also bring those vendors to a halt instead. As Weinman said himself, users may be forced to `pull from the open source world’, to which the short response is `yes….and?’ Why not source applications and services from the open source world?  

One answer is that suitable code is not necessarily available, but this is where the current economic situation, coupled to the disruptive potential of The Cloud, may re-run old historical patterns. The Cloud most certainly represents a brand new marketplace with totally different dynamics, occurring at a time when many users will be putting current investment plans on hold for a while. 

That gives a significant opportunity for new start-up companies that do `get’ how The Cloud’ can be exploited to develop their technologies and business models. When users start to see their way clear for more concerted investment plans they will also be in the market for whatever maps best onto their needs – which are likely to encompass much greater flexibility and agility, far greater performance, lower maintenance cost through self-management and healing, and dramatically reduced energy consumption – with much of it available in practice as a pay-per-use service. 

This is likely to be beyond the scope of many of the currently leading ISVs, particularly if they stay with the old licence models. The start-ups are also likely to have far better technologies which hit the new sweet-spots that post-slump businesses will need. After all, a cynic might actually wonder why all the current crop of fancy financial risk analysis tools were unable to spot that large packages of dodgy debts were still very dodgy debts, despite being rebranded as `investment vehicles’.

When it comes to IT, the word `monopoly’ is invariably heavily loaded with portents of controlled availability and inflexible product offerings. This is certainly the perception that comes hand in glove with IBM and the mainframe marketplace.  

The truth about any monopoly, however, is that it has the potential to be at least as good for users as any perceived badness normally associated with the word. And with the recently announced acquisition of small, Sunnyvale-based rival, Platform Solutions Inc (PSI) IBM now gets the chance to demonstrate – under a very public magnifying glass – just which interpretation of `monopoly’ it is going to exercise. 

To call PSI a `rival’ to IBM in the mainframe marketplace was, of course, stretching the word to near breaking point, but its technology – a hypervisor that allows mainframe code to run on Itanium-based hardware – has (or possibly had) the capability of opening up the mainframe market outside its normal user base. Its key advantage is to offer the possibility of a lower-cost, probably lower-performing alternative platform. This could help move mainframe applications into the mainstream of event-driven web services, for example, and could run them on the same platform as a wide range of applications from other platforms. 

There were, of course, the crossed lawsuits that existed between them: IBM suing for copyright infringements and PSI countering with an anti-trust suit; but they are now gone. It would be easy to assume that IBM bought PSI just to stop the action against it, though I suspect its lawyers could have kept the legal wrangling going long past PSI’s ability to fund the fight, and IBM’s lawyers still need paying, regardless. Such an action would also be a certain mark that company favoured the `bad’ side of monopoly. 

That fear has been expressed by US Computer & Communications Industry Association president, Ed Black. He has suggested that “it is transforming a market with latent potential for competition and innovation into a sector with little prospects for anything but complete domination of IBM.”

Well, yes, that potential is now stronger than ever. But IBM has a history of allowing different divisions to compete for the same markets, and sometimes compete with extreme prejudice. In the mainframe market, where the development costs in both hardware and software (and systems software in particular) can be mind-bogglingly high, monopoly may in practice be the only way to make it work. But that does not mean the only answer needs be z/Series, and only z/Series. IBM can have an over-arching policy for, what shall we call it?, big enterprise solutions, but will happily let the divisions fight it out for a share of the business.

The company has already hinted that the PSI technology is set to be exploited for future products, though they are unlikely to appear as z/Series systems. One way (and potentially a very charitable one) this can be interpreted is as a pitch at creating a win:win:win situation.  

The market would win because it would get delivered, from a different division of IBM, the types of platform that do accommodate the mixing and interoperation of radically different operating environments. It could also open up the market for new users that could make business sense of a smaller, cheaper, slower (but just as robust and secure) `mainframe’ system. Here lies real market potential, but arguably one the z/Series team might not feel happy with. And as this is a software technology running on a standard platform, other vendors could be licenced to sell it too. 

Third party applications and tools vendors would win because new business opportunities would be opened up for them. It could also open up the market to competition from ISVs from outside the mainframe community. 

IBM would win because it would maintain control over The Mainframe at a strategic level. Despite the inevitable cries about stifled innovation, the upfront costs of playing in the mainframe market – either as vendor or user – are such that a good deal of strategic control is necessary. Innovation can be a good thing, but it can also kill.  

Let’s not forget that the wild-frontier web services sector actually exists off the back of a raft of standards that vendors actually adhere to. Thankfully, the days are now gone when the likes of dear old Ken Olsen, founder of Digital Equipment, would utter sayings such as (and I paraphrase) `we believe in the standardisation of Unix, and will differentiate ours by including this, that and the other that nobody else can offer’. 

And yes, this is a charitable view, and it is IBM. We must wait and see. But don’t cry for PSI, for its legacy may yet be marked by important changes.