Sometimes, when I'm sick of the hype curve, I wish people would stop selling technology before it is ready. Just keep it in the labs till the standards are fully defined, and we know the products will work. It works (well, more or less) for the pharmaceutical industry.
Of course it could never happen in IT. Apart from anything else, products and technologies get defined and redefined by users trying out the early versions. IT is a heuristic process. It's not possible to do that in drug development, but in IT, we get to fix things as we go along.
If IT managers had to operate with stuff developed in isolation from the market and from users, we'd be stuck with white elephants like OSI, and probably still be using primitive 70s-style mainframes.
So, when it comes to Web services, we need to buckle down, and accept that there are going to be serious bumps between here and the promised land. But there are limits -- if things get too bumpy, significant numbers of IT managers will decide that the benefits aren't yet worth the effort.
Part of the payoff is that we get to see more creativity from smaller players. IT is not (yet) quite as dominated by giant companies as the drugs industry. "IBM, Sun, Microsoft and BEA are way behind the smaller guys," says David Sprott, chief executive of analyst from CBDI.
From the user side, it seems as if IT managers are prepared to accept the way things are at the moment -- Web services adoption is going pretty well, according to a study from CBDI. It looks as if users are finding that they can save money with Web services -- there is pretty much no other reason why they would do it at this time. Those cost-savings, by the way, are mostly in allowing users to link together incompatible legacy systems, instead of replacing them.
However, the heat in the standards kitchen does seem to be getting a bit fierce, with Microsoft and IBM backing out of a World Wide Web Consortium (W3C) standards process in favour of its own answer, in one of those standards spats that can blight a technology's early years.
The issue is what is referred to as choreography, or orchestration; harmonious-sounding words for something that has become quite acrimonious -- by the standards of standards makers. "Orchestration has got something in common with grid computing," says Roman Stanek, chief executive of Systinet, a smaller Web services player just expanding into Europe. "An orchestration product will work like a message broker linking Web services."
This is not a big issue yet, since most Web services implementations so far are operating more or less in isolation. Orchestrating or choreographing them is not a problem. "Orchestration will become interesting, but it will take time," says Sprott. But it looks as if the process of creating a standard could damage the credibility of Web services big-time.
Microsoft and IBM have had a look at what the World Wide Web Consortium is doing in the area -- an effort called WSCI (Web Services Choreography Interface) and backed off, created their own Business Process Execution Language for Web Services (BPEL4WS -- or BPEL for short), jointly with BEA, and -- so far -- not put it to any standards body.
Standardisers have expressed a fear that these big guys want to charge royalties, something they could not do within W3C rules. But IBM has denied this. It seems far more likely to me that they simply want to make sure they define the standard. "IBM and Microsoft want a hegemony to drive the major standards," says Sprott. "That is absolutely crystal clear."
In his view, IBM is joining forces with Microsoft to make sure that other players (and particularly Sun) don't get to define the Web services playing field.
I can see two problems with this. Firstly, a specification defined outside of an open process will very likely not be as good as one that gets more scrutiny. I may be completely wrong here, as even Stanek -- and his Systinet is one of the little companies whose interests W3C is in theory protecting -- thinks that W3C is on the wrong course, and becoming more and more irrelevant in comparison with the XML standards group Oasis, and the Web Services Interoperability (WS-I) group that just weighed in on security.
Now, the big guys behind BPEL could stifle some criticism by pushing it through one of these supposedly more switched-on bodies, but that would not solve the problem -- WCSI and BPEL would then be rival standards, something which would kill most users' enthusiasm for adopting either.
Secondly, a delay in standards will not result in any delay in products reaching market. This is simply because vendors need to show some revenue. "Many of these companies have large venture capital VC investments, so they are under huge pressure to meet their VC funds' targets," says Sprott.
Systinet, for one won't wait to get into orchestration, and others such as Collaxa, Intalio and Choreology will be in there too. Intalio, for one, has people on the W3C's WCSI group. What these people do may be collaborative, says Stanek, who hints that he may work with one of them to offer something on orchestration next year. "It should be very interesting."
So, immature and disputed standards will be pushed to the market by companies that need to make sales to stay alive. And users who see short term savings and respond to the marketing pitch may find that they have built yet another incompatible silo system. The level of incompatibility will be much smaller than between their legacy distributed systems from the 1980s and 1990s. But it is still ironic, because incompatilbity is exactly the problem Web services was supposed to solve.