to homepage
 Weekly emails: how to advanced search
 Glossary lookup:


Loosely Coupled weblog

Friday, May 10, 2002

Composition versus orchestration
I watched a demo of WSFL in action yesterday while sitting alongside chair Howard Smith, whose perspective was both well-informed and thought-provoking.

Web Services Flow Language [PDF file] (WSFL) has been proposed by IBM as a specification for modeling business processes when using web services to create or integrate applications. The Business Process Management Initiative ( was formed in August 2000. Just as there is a relational database model that underpins today's RDBMSs (relational database management systems), expects that its language for managing business processes (BPML) will enable BPMSs — Business Process Management Systems.

At first glance, WSFL and BPML sound almost identical, except of course that modeling is not quite the same as management. In fact, the difference is a chasm.

Yesterday's demonstration was provided by SilverStream, which in March released a new version of eXtend Composer, its J2EE-based XML integration server. One of the new features in that release was a visual process modeling engine based on WSFL, billed by SilverStream as the first commercially available business process manager based on the IBM specification. That component formed the core of yesterday's very impressive demonstration.

The full stack of SilverStream tools adds up to a complete web services development and deployment environment that has very few peers in today's market. It is the complete environment that IBM's WebSphere platform family is on its way to becoming, and which BEA would like to have. Iona comes closest with its Orbix offering, but SilverStream adds the humility that comes from being dominant in none of its markets, which gives it the edge in adaptability. It is a marvel to watch the eXtend suite in action, making complex B2B and EAI integration tasks appear almost effortless. In his day job, Howard Smith is European CTO of systems integration titan CSC, and I put it to him that development environments like SilverStream are automating a lot of the integration grunt work that used to provide bread-and-butter revenue for the IT services industry. "This is a smarter form of systems integration which lets customers and their IT partners focus more on the business processes and less on the technical integration," he said, which I took as tacit acknowledgement.

Yet watching SilverStream demonstrate the ease with which processes could be defined, manipulated and automated using its environment, I still couldn't throw off the sensation of concrete being slowly poured into the business infrastructure. Sure, it was being poured more effectively and speedily than I've ever seen it poured before by a technology vendor, and in a way that meant making changes later — like pulling down a partition here or adding an extra doorway there — was rarely more than a day's effort, and could never threaten the integrity of the entire software infrastructure. But as Howard pointed out, it was still starting from the premise of predefined workflows, where each participant's role was mapped out in advance. His verdict: "We have to be careful not to embed our processes in the way we integrate, just as we've found that packaged applications ingrain the business logic."

BPML, in contrast, makes no assumption of a grand plan that has already mapped out the roles of every participating component. Instead of imposing software constructs on business processes, it treats them as what they are — as quasi-organic entities, each with their own autonomous identities and delegated freedom of action. As a way of helping to get your head around the concept, imagine business processes as a community of weblogs, each one self-contained and independently free to define how it fulfils its purpose, yet still capable of being harnessed to produce a pooled result.

Howard is quick to point out that this is more than a vision. CSC is implementing this form of process management in production environments for real customers. "We now know how to do loosely coupled, massively distributed interaction, without any centralized control," he told me yesterday.

The catch of course is that it will cost you a lot of money to have CSC implement this for you if you want it today. The majority of enterprises are still stuck with SilverStream and its ilk — and for most of them, that will be a very good thing indeed, when you consider what they had to work with before. Just bear in mind that it's not the final destination yet. There's still a world of difference between composing workflow and orchestrating business processes. Even the word orchestration probably understates what we need to achieve here. This is an orchestra where every player is improvising to a different score, where multiple conductors each set their own tempo, in a performance where every listener hears a different tune, and yet somehow a shared awareness of the underlying harmonies ensure that every strand still sounds melodious.

PS: There are currently four main web services specifications for business process modeling and management. There's no really definitive resource available online that compares them, but to get snapshot of the runners from the point of view of one industry participant, see this recent article at Web Services Architect. For regular news on business process management matters, see this weblog maintained by Howard Smith along with Gillian Taylor and Peter Fingar.
posted by Phil Wainewright 8:26 AM (GMT) | comments | link
Location neutral, network resident
Although it's early days, we can start to identify the key architecture elements for loosely coupled computing:
  • A shared, seamless, always-on infrastructure

  • Platform independence, on both the server and the client

  • Distributed autonomy, including support for local disconnected computing

  • Network-resident management (as distinct from centralized control)

This topic requires much more space to expand on these concepts than a weblog entry allows, but I was minded to jot them down while thinking some more about the Macromedia MX architecture I mentioned previously. Despite various shortcomings in the execution, conceptually it maps quite well to the above points, which is why I warmed to it. Jon Udell touches on similar ground in this erudite review of MX: "The Flash run time is already positioned as a universal Internet client that can connect to distributed Web services — a core benefit that [Microsoft's] CLR may not achieve for years ... If Macromedia can transform Flash from a multimedia shell into a popular presentation layer for Internet-based business applications, the missing pieces of the web services puzzle — identity management; security context; reliable messaging; and real-time, two-way data communication — will come into focus more quickly."

But despite the undoubted talents of its development team, there is an unfortunate credibility gap that Macromedia has to overcome, which is not entirely of its own making. I am one of the many web users who inwardly groan each time they see the words, 'Flash loading'. In the minds of a significant number of business people — the constituency Macromedia must now target with its MX offering — Flash has become synonymous with costly, ineffective, vainglorious website design. The manner of its previous success has created a brand image that is completely opposite to the virtues it must now promote if it is to gain success as an application platform.
posted by Phil Wainewright 4:01 AM (GMT) | comments | link
Digital identity is key to service architectures
Microsoft's decision to package Passport as an enterprise product was a necessary sacrifice to ensure enterprise adoption. Charles Fitzgerald, general manager of Microsoft's .Net platform strategies, explains the background in this interview with Digital ID World: "We think the concept of user-centric web services, of anytime, anywhere, any device access to data is as compelling to the enterprise as it is to a consumer ... the level of demand from enterprises to run their own instances of My Services inside the firewall ... frankly ... came as a surprise to us because we saw a lot of the value coming beyond the firewall."

Microsoft realizes that in a highly distributed, loosely coupled service architecture, access security can only be enforced at the level of the individual user. Relying on physical boundaries and firewalls alone won't be enough. That truth is also recognized in the very existence of a site like Digital ID World, whose strapline quite rightly asserts: "Identity is center".

"The challenge now is to hook up all the disparate identity systems so that there is more value and convenience for business users and consumers across a much wider network of services," said Fitzgerald later on in the interview. It's true that having Passport operated by a single entity would have been a neat solution to that challenge, but it proved unacceptable in practise. Hence the importance of formulating standards like WS-Security.
posted by Phil Wainewright 3:09 AM (GMT) | comments | link

Thursday, May 09, 2002

The web is now the default platform for applications
If Gartner says so, it must be true: "All new applications are going to be web-based, regardless of how they are going to be deployed," according to Massimo Pezzini, VP and research director of application integration and middleware strategies at Gartner Group. He was speaking earlier today in London, at a seminar laid on by web services platform vendor SilverStream that provided much food for thought (more on that tomorrow).

He cited several factors that sustain the inevitability of the web as the default platform for applications both within and beyond the enterprise:
  • "It's almost impossible to find a development tool that is not web-based." So even if you wanted to develop some other way, you can't.

  • "Whatever new platform you are buying, web technology is there and increasingly web services technology is there. Web services is becoming a commodity." It's not an option that you pay for, it's automatically built into any software infrastructure products you buy, as standard (including mainframe products like CICS and IMS).

  • Web services is a new and better way of doing everything that people were already trying to do with CORBA, DCOM and the like. "The importance of web services from an architectural perspective is that web services are leading enterprise users to adopt a service oriented architecture."

One other interesting point about web services is that no one vendor can own the architecture, according to Pezzini: "The reality is that web services are by definition platform neutral."

He mentioned a load of other good stuff about 'spaghetti code' and composite applications, and how you move from one to another, but those points will have to wait for another day.
posted by Phil Wainewright 1:45 PM (GMT) | comments | link

Wednesday, May 08, 2002

Macromedia goes loosely coupled
Macromedia's new MX architecture is an imaginative pooling of its client tools with Allaire's ColdFusion platform, acquired last year. This interview with company CTO (and Allaire founder) Jeremy Allaire gives some background on the forward-looking thinking behind MX, which is designed to embrace the jump towards participation on the Web: "According to Allaire, people want to start actively doing things on the web, as well as just reading documents ..." Macromedia is positioning its new Flash MX client as a platform for rich client capabilities, able to interact with and invoke components running on the Cold Fusion MX server, which can now sit on top of J2EE platforms such as WebSphere or WebLogic as well as Macromedia's own.

Rather like Microsoft's Visual Studio .Net, MX is going to give a lot of developers a sudden and unexpected introduction to the web services age. It will however have to do more to make its Flash environment accessible to non-developers if it is really going to make waves with MX.
posted by Phil Wainewright 1:53 AM (GMT) | comments | link
Weblogs and journalism
More advice for journalists from Dave Winer, this time on how to start a weblog. It's good advice for a journalist (or indeed for any professional) who wants to use a weblog as a way of keeping contacts informed on what they're working on, and of giving a detailed insight into their background, experience, interests, lifestyle and personality. But that's just one way of using a weblog.

One of the core skills that people turn to journalists for is their ability to identify and home in on what's important in a topic. This site aims to be a resource for people who are interested in on-demand web services, and I don't believe that what I saw at the cinema last night is relevant to that. This is a different kind of weblog, one that revolves around a topic (along with my perspective on it) rather than an individual person. In fact, I think of it as more of a k-log, a concept that Dave's colleague John Robb is keen to promote.

Nevertheless, there are some important weblog elements Dave mentions in his piece that I still need to add to this site, which is why this page is still headed up as "pre-launch". Creating the full set of features is a voyage of discovery, one that I am greatly enjoying and which I hope, in some small way, will make a contribution to the continuing evolution of the weblog as a vehicle for writing on the web.
posted by Phil Wainewright 1:35 AM (GMT) | comments | link

Tuesday, May 07, 2002

The role of professional journalists
Dave Winer has been complaining that professional journalists aren't taking any notice of what's happening in blogland: "The professional journalists have been asleep at the wheel, leaving a void."

But Dave is confusing professional journalists with commercial publishers. A good journalist is someone who can find a topical subject and write about it in an interesting way. A successful journalist is someone who then sells that story to a publisher. Something that I learnt when I founded in late 1998 was that mainstream titles don't cover a topic until it becomes mainstream (the same, by the way, is true of analysts). So emerging stories get covered only sporadically, and then mostly without any comprehension. That's not the fault of the journalists, it's the way the system works — and in the end, it comes down to what the readers want to read about. (Here, by the way, is what in my professional judgement the readers of UK computer trade mag MicroScope would want to know about weblogs two weeks ago; way less than I really wanted to say on the topic, but for most of those readers it would be their first exposure to the concept).

Once I'd written all the articles I was likely to sell about ASPs back in September 1998, I went off and founded my own website so that I could continue to write about a topic that I knew was important. Having sold that site 15 months later, I can now afford to found this website to write about another topic that I know to be important. Some of us have noticed, Dave, but it will take a while longer before the mainstream gets it.

posted by Phil Wainewright 3:28 PM (GMT) | comments | link

Monday, May 06, 2002

From project-centric to process-centric development
Commenting on BEA's Weblogic Workship (formerly known as Cajun), web services startup Collaxa made an interesting point in a profile in this month's Stencil Group newsletter: "Workshop is a rapid services creation, not orchestration, tool."

What this comment reveals is that the project-centric model of creating an application and then deploying it is so ingrained that even a switched-on vendor like BEA is still designing its tools for that old model. That leaves plenty of room for new-wave vendors such as Collaxa to come in and sell tools that support the constant evolution and adaptation of business processes. BEA is still mired in the business of selling concrete. Collaxa and vendors like it provide the tools to orchestrate malleable web services putty.

With founders drawn from the Netscape and NetDynamics teams that created two of the most influential first-generation application server products, Collaxa is clearly a company worth watching.
posted by Phil Wainewright 2:48 PM (GMT) | comments | link
Focus on your passion, outsource everything else
Last week Userland had a major outage that temporarily disconnected several key elements in the global blogging grid:
  •, the Userland-operated site that shows a running list of recently updated weblogs, was down and therefore inaccessible unless you could find the mirror site that was hastily put up as a temporary measure

  • Posting to Userland-hosted weblogs from Radio, the company's ground-breaking weblog client, was temporarily suspended (though Radio does allow users to host elsewhere, so those with the technical know-how and resources could bypass the problem)

  • Userland CEO Dave Winer's prodigious output of content across multiple weblogs was significantly diminished (though far from unbowed)

One comment from Dave Winer in the midst of all this is worth highlighting I think:
We are software developers, not system operators ... When we get the system working again, we're going to be looking for one or more partnerships that allow us to offload big pieces of the system ops. There are organizations and people who have much more knowledge about this. We love to develop software, and we're really good at that. That's where UserLand's focus should be.

The shared infrastructure of the web means that no-one needs to stray outside of their core competence (their passion) any longer.

Coincidentally I was browsing Dan Bricklin's weblog at the weekend and came across a link to this article about how the AmIHotOrNot website was set up. This was the site where teenagers rated their peers' pictures for attractiveness (heat?) and could submit their own for similar treatment, which became a raging overnight success early last year. The site's founders — who are also responsible for the rather more serious web services directory — didn't know much about managed hosting or distributed services when they started out, but they had the sense not only to host the site with the ultra-reliable Rackspace Managed Hosting service but also to get users to host their image files on hosting space of their own. In fact, that decision not only saved the venture from collapsing under an unbearable bandwidth overload, it even ended up generating revenue: "... it occurred to me that some companies might actually want to host peoples' photos and pay us a bounty for sending them users. By directing our users to these companies, we turned one of our major costs into a revenue stream," notes founder James Hong.
posted by Phil Wainewright 1:16 PM (GMT) | comments | link

Assembling on-demand services to automate business, commerce, and the sharing of knowledge

read an RSS feed from this weblog



latest stories RSS source

Headline news RSS source


Copyright © 2002-2005, Procullux Media Ltd. All Rights Reserved.