It's nice to see Scott McNealy and Bill Gates agreeing on something. Sun's CEO was in London yesterday, saying that all computing now runs on web services ("It's so last millennium to write to the OS"), reports The Register. Meanwhile, Microsoft's chief architect was telling financial analysts in Seattle this week that web services will be "built in" to the next major release of Windows, codenamed Longhorn.
Evidently both these visionaries believe that future applications will be written to take advantage of web services rather than the underlying operating systems. But inevitably, their agreement runs only interface-deep. Below the web services API, Gates is backing the Windows-native .NET environment, while McNealy plumps for Java web services ("open, runs on every OS").
What neither of them mentioned is that the beauty of web services, of course, is that you can have both, and they interoperate. Scott, unfortunately, has a "last-millenium" vision of his own, which is that somehow the world would be a better place if all the computing power were somewhere "out there" in the network (with the hidden subtext of "running on Sun hardware"). The point this vision misses about the network is that it isn't a consolidated, centralized backbone affair. It's distributed all around us, and in many cases it makes a lot more sense to put the power near the users rather than transporting it away to some arbitrary remote location. Web services lets you do that, using .NET and Windows if it's appropriate, as well as allowing you to have the high-powered centralized boxes for those functions that make sense as consolidated shared services.
Another of Scott's redundant notions is his comparison of computing with automobiles: "You do not, says Scott, buy the bits then build your own car, and it will become similarly absurd to buy a piece of hardware then add the software yourself." The analogy fails because it doesn't take into account the network context of each of those products. Cars are no use without roads, and people buy cars so that they can have the freedom to drive where they like. Similarly, computers are no use without software, and buying a computer with all its software built in is as absurd as buying a car that will only run on its own private road network.
People buy computing so they can have the freedom to automate the things that they need to do faster, cheaper or better. In the same way that a car driver plans the best route to their destination each time they get in a car, so a computer user wants to assemble the best automated services available for the tasks they want to complete each time they turn on the computer. The significance of web services is that the technology is going to empower ordinary users to do that everyday service configuration task, instead of their having to rely on software engineers to do it for them.
posted by Phil Wainewright 6:52 AM (GMT) | comments | link
Thursday, July 31, 2003
After predicting Siebel's impending demise last week, I am astounded to discover that company founder and CEO Tom Siebel has just funded a dinosaur museum to the tune of $2 million. CNET reports that the eponymously named Siebel Dinosaur Complex (no, seriously) at the Museum of the Rockies in Bozeman, Montana, is the recipient of a $2 million personal donation from Tom and his wife.
Of course, what company executives do with their own money in their own spare time has no direct connection with the businesses they run. But if my prognosis for the company is anywhere near right, then Tom Siebel's sudden decision to invest in educating the public about what happened to the dinosaurs will soon turn out to be a monumental example of unintentionally self-inflicted parody.
This is a pattern that I've noticed recurring time and again when individuals or institutions pass their all-time peak. Still in denial of their impending decline, yet subconsciously aware that the good times are over, they come out with some gesture that's intended to reinforce their good fortune, and yet by some twist of fate it ends up as a hollow satire of their former glory.
I have to confess I was somewhat distressed earlier this week to read about the Pentagon's extraordinary plan to set up a derivatives market to help it assess geopolitical risk. The astonishing conceit that the Policy Analysis Market would somehow be helpful in predicting future shocks seems to me to be such a striking example of self-parody that it immediately makes me fear for the stability of the multi-trillion dollar global derivatives industry. That otherwise perfectly intelligent and well-educated people are still arguing that it would have been a good idea demonstrates such naive belief in the power of derivatives markets that I'm afraid some rude awakening lies just around the corner.
Back in March 2000, I wrote an article satirizing the dot-com mania of the time, little suspecting that just a few months later, right on the cusp of the subsequent crash, VCs would be actually be funding real-life versions of the idea I'd satirized.
If markets were indeed a reliable guide to the future for those who analyze their behavior, then the dot-com crash wouldn't have caught quite so many VCs and fund managers (let alone individual investors) on the hop. The only good guide they provide is as a barometer of complacency and self-delusion.
So too with individual acts. The ancient Greeks called such acts hubris, which brought down the wrath of the gods on those who committed them. In these more enlightened days, I prefer to pin the blame on arrogance, denial and self-delusion. But whatever the causes, the results are the same. If you or your company are heading into an endeavour that could end up as unintended self-parody, in my view that's a sure sign that things are not as sound under the surface as you've convinced yourself they are.
PS: Tom Siebel praised the Montana museum for how user-friendly it had made its exhibits, reports the Bozeman Daily Chronicle. The museum has appointed its star paleontologist, Jack Horner, to design the Siebel complex, which will take up approximately 12,000 square feet. "But beyond the size, details are sketchy," reports the paper, apparently without irony.
posted by Phil Wainewright 7:01 AM (GMT) | comments | link
Our happy community of web services-oriented bloggers has just woken up to the fact that Adam Bosworth started a weblog last weekend. This is a welcome development on two fronts:
Adam's weblog is already stimulating debate on important topics in the development of the services architecture;
Having such an influential figure keeping a blog is a futher validation of the medium.
Adam, of course, is chief architect at BEA Systems and previously worked at Microsoft, where he was a key figure in the early development of the first web services standards including SOAP. He links from his weblog to a March interview in ACM Queue magazine that gives a useful update on his current thinking on web services.
For those readers of Loosely Coupled who haven't yet acquired an RSS reader (or who prefer not to), I've taken the opportunity to add Adam's weblog to the short sampling of web services-related weblogs on our news aggregator page. I've also added some other notable blogs that are making important contributions to the debate on services architectures. This is far short of being an exhaustive list, and it would be better if it were organized chronologically instead of by blog, but it assembles some of the key voices for the convenience of those who don't feel like tracking every single contribution to the debate.
posted by Phil Wainewright 4:48 AM (GMT) | comments | link
Tuesday, July 29, 2003
Just when you thought no one could possibly dream up yet another web services standards proposal, along come three at once. The Web Services Composite Applications Framework (WS-CAF) comprises three separate specifications that aim to co-ordinate transactions across multistep processes in a B2B environment.
Sun's name at the head of the WS-CAF proposal has prompted the usualcrop of kneejerk outbursts that pass for news coverage of web services standards stories in the technology media these days. Only eWeek maintains a balanced perspective in its coverage. The rest rely heavily on what appears to be a round-robin statement from ZapThink analyst Ron Schmelzer that having yet another competing standard is a bad thing. (Somehow, he manages to give identical quotes to two entirely different publications belonging to competing publishers).
As I've said previously, we should celebrate rather than decrying the emergence of competing specifications particulary when each specification addresses a separate need. Ron himself admits that WS-CAF deals with electronic transactions that run across separate organizations, and therefore has a different emphasis from the IBM/Microsoft-sponsored BPEL initiative, with its WS-Coordination and WS-Transactions specifications.
The most important distinction is between directed orchestration, which lie at the heart of the IBM/Microsoft specifications, and autonomous choreography, which is the province of the WS-CAF proposals. The IBM/Microsoft technologies are designed for scenarios where someone's in charge, typically within a single organization or a tightly managed collaboration. But in a complex B2B trading environment, there's no single point of control with that degree of authority over the participants, and therefore co-ordinating processes is the subject of a great deal more negotiation between independent parties. The task of maintaining transactional integrity in that environment is not something that can just be tacked on to the IBM/Microsoft specifications as an afterthought. Hence the need for WS-CAF.
Those who enjoy seeking out standards dissent don't have to be disappointed, though, because WS-CAF overlaps much more strongly with the existing OASIS standard BTP, originally designed to govern transactions in an ebXML environment, and adopted by London-based vendor Choreology as the basis for its business process co-ordination platform. Choreology which was one of the companies featured in our recent article on orchestration and choreography, Bringing it all together was founded by former members of HP's Arjuna Labs.
When HP closed down its Bluestone middleware division, another company called Arjuna Technologies was formed from the remains of the Arjuna Labs. Based in Newcastle, in the north of England, Arjuna also has a web services product based on BTP, but has now become more notable as one of the leading co-authors of the WS-CAF specification.
You can imagine that the inevitable rivalry between these former colleagues potentially makes a much more interesting standards-battle story than the hackneyed old Sun-and-Oracle versus IBM/Microsoft one. But to be honest, whatever rivalry there is between the Choreology and Arjuna teams is probably not bitter enough to make the front pages of the tabloids. There's enough scope in the complex arena of B2B transactions for there to be a need for two separate specification layers.
Now that all these proposals are out in the open, what needs to happen next is a mature discussion that leads to the important distinctions being preserved, while any duplication can be eliminated. Fortunately, that process is unlikely to be sidetracked by a handful of inadequately-researched news stories based on opportunistic soundbites.
PS: Some laudably calm counsel comes from Doug Kaye on the topic of web services standards: "Unlike the monolithic specifications of the past, the WS standards bite off smaller chunks of the problem at one time. They're 'composable' (able to be combined) in the same way as web services themselves. Is this modularity of specifications as chaotic as it appears? I think not."
posted by Phil Wainewright 12:06 PM (GMT) | comments | link
The article sets out the most common styles of buying web services among enterprises today:
Inside-out, where web services are first used internally as an alternative to enterprise application integration, and then extended later on to reach out externally to trading partners. This is most common in industries that have complex legacy IT infrastructures, he says.
Outside-in, which starts out by deploying collaborative web services at the edge of the enterprise in customer-facing or supplier-facing business processes. Industries including electronics, automotive, and insurance are "more prone" to this form of collaboration, he writes.
There's nothing particularly new, here, of course; and he also notes that many organizations are doing both at the same time. It gets more interesting when he goes on to discuss some of the possible motivations driving web services adoption (or the lack of it):
Play it safe: The key motive here is to minimize risk, and inevitably it leads to limited, behind-the-firewall implementations.
Business value optimization: This approach targets web services deployment, even of pilot projects, where business value can best be realized. This is Eric's favorite (and mine): "Web services are a business decision first, followed by the technology decisions."
Modified tire kicking: Here, Eric recommends spending a lot of time asking tough questions of a broad spectrum of vendors not to waste their time, but to quantify and assess the risks.
Out of all this, Eric distills a list of six recommendations for buyers embarking on a web services project, with which he concludes the article. They neatly tie together all the issues raised in his earlier points, and will help fortify the resolve of anyone hesitating about acting on his final words of advice: "Get started with web services now."
posted by Phil Wainewright 3:58 AM (GMT) | comments | link
Assembling on-demand services to automate business, commerce, and the sharing of knowledge