IBM and Microsoft have pooled their existing business process languages to create a new joint specification, Business Process Execution Language for Web Services (BPEL4WS), which they will announce this coming Monday, joined by BEA Systems.
According to a report in Network World Fusion, BPEL is "a workflow language that describes the number of web services that need to be executed, the order in which they are executed and the type of data they share." It works in tandem with two other specifications that will also be unveiled on Monday. WS-Coordination conducts the interaction between the web services that make up an end-to-end process, while WS-Transaction monitors the process to ensure that all the required transactions have been fulfilled before confirming it as completed (additionalreporting from CNet and InfoWorld respectively).
"The three specifications will work in harmony to solve the workflow issues involved in connecting and executing a number of Web services that may be running on disparate platforms internally or across multiple enterprises especially in an electronic commerce scenario," summarizes Network World Fusion's report.
The inclusion of WS-Coordination appears to sideline WSCI (Web Services Choreography Interface), the specification announced in June by BEA, Sun and Intalio, which only this week was officially recognized (though not yet endorsed) by the Web standards body W3C. There has been no official response as yet from BPMI.org, whose Business Process Modeling Language developed by Intalio may also be overshadowed by BPEL.
"BPM provides the ability to aggregate tasks/processes; determine state, audit, manage service-level agreements; and escalate in fact, all the missing components in the Web services world today ... Together, both of these layers can arguably produce the fastest and greatest ROI of any other technology combination today ... Web services offer BPM suppliers a new way to build business applications. Applications can be built from a process perspective, with the process itself being central to the application."
Wrapping objects in web services is a bad idea, writes Adam Bosworth, an early champion of XML at Microsoft and now VP engineering at BEA Systems, in an article in XML & Web Services Magazine:
Objects are repositories of state. Conversations with them are by definition not stateless. Because objects are encapsulated, conversations with them are also inherently fine-grained ... Web services doesn't mean surfacing application "interfaces" to underlying objects through automatically generated SOAP. It means providing well-defined, coarse-grained messages that provide all possible information in one fell swoop (SOAP) and a contract (WSDL) for which messages sent in result in which messages sent back.
Bosworth has consistenty maintained that coarse-grained messaging is a fundamental tenet of a successful web services architecture. This article is a succinct, clear and accessible explanation of why that is even down to its title, One Fell Swoop, which is a catchy alternative way of saying 'coarse-grained'.
Ed Khodabakchian adds an interesting take on this in the Collaxa blog[NB: URL updated Aug 10]: "Think of Web Services as queues that consume and generate XML Schema Typed messages. That will force your design to be coarse-grain and loosely-coupled."
But if you're not allowed to maintain state, don't you end up with a meaningless cacophony of disconnected statements? Sean McGrath raised this question in a neat article entitled A Statement is Not a Conversation in last week's XML in Practice newsletter from XML.org: "No amount of beautifully crafted statements adds up to a conversation. A conversation requires ebb and flow, the passage of time, you need a dance...."
Of course, that's where orchestration (or, more in keeping with Sean's metaphor, choreography) comes in. It's also why Adam Bosworth stresses asynchronous messaging and a loosely coupled architecture as the two other pillars of web services. These acknowledge the need for a conversation, but they also emphasize that it is a conversational style that has to be tolerant of interruptions, requests for clarification, delayed responses and seemingly irrelevant digressions.
posted by Phil Wainewright 1:44 AM (GMT) | comments | link
Thursday, August 08, 2002
Blogging and the killer app
Dan Bricklin's company Trellix this morning announced it has added blogging capabilities, based on Blogger technology, to its website building software. This is the first mainstream website management tool to integrate weblog capabilities 'out of the box' a term that is strictly metaphorical when applied to Trellix, since its website builder is a hosted platform, accessed via a browser. Because it provides its software on a private-label basis, Trellix is not well-known as a company, but through customers including BizLand, CNET, Domain Direct, Interland, McGraw-Hill, Namezero, Terra Lycos and Thomas Publishing, its platform has launched millions of websites. In terms of market share, positioning and substance, Trellix is the FrontPage of hosted website publishing.
Trellix targets small businesses, so this announcement is a first from several angles:
The first mainstream website builder to offer blogging as a standard feature
The first company to market blogging explicitly as a business tool
The first blogging tool with a full set of complementary website building functions
That last point addresses one of the biggest weaknesses of blogging software to date. The leading weblog platforms are great if all you want to do is to publish a blog. But if you want to add other pages or functions, or integrate the blog within an existing site, then you have to be something of a website wizard to do it. Indeed, as I've mentioned previously, Blogger's rival Userland actually puts obstacles in the way of cleanly integrating into the look-and-feel of an existing site something the company will have to rectify if it wants to make the headway in the enterprise market that it covets so keenly.
The people at Trellix have been thinking quite carefully about the business applications of blogging, as Dan makes clear in an article posted to his own weblog site this morning:
"Besides links, opinion, and expertise, we also see how a blogging component to a web site authoring system can be used for other applications, including change logs, on-going status, and more. From a small business and organizational perspective, this is an area that has lots of fertile ground."
We'll have to wait and see what it is that businesses choose to do with their Trellix weblogs, because today's announcement marks availability of weblog capability to the company's private-label customers, none of whom as yet have chosen to implement the option. But Dan notes that "potential customers [are] starting to specify blogging as a requirement to some deals, and a highly desirable feature of others," so it shouldn't be long before the first of them goes live.
Dan Bricklin's involvement is in itself another factor that makes today's announcement specially notable. Dan is something of a computing industry icon, best known for his role as the creator of the PC spreadsheet and founder of VisiCalc. By allowing accountants and business analysts to easily automate calculations they had previously had to perform laboriously by hand, the spreadsheet became the 'killer app' that launched the PC into the business market. For many of these early adopters, the Apple IIs and later on the IBM PCs on which the software came loaded were little more than the packaging surrounding the spreadsheet. Without the 'killer app' functionality delivered by VisiCalc and its later rival Lotus 1-2-3, the PC revolution might never have happened. Thanks to Dan, it did, and the rest quite literally is history.
Is today's announcement a sign that weblogging is the 'killer app' of web publishing? Personally, I don't think so, but that's only because I think the killer app is something that's much bigger than blogging alone. Blogging has been a vital first step it's become a powerful demonstration that ordinary web users can be directly in control of publishing their own content online. But the real killer app is the ability to directly control both content and functionality.
An individual web site (come to think of it, the Web itself) is just like a spreadsheet, full of blank cells that can be filled either with content or functions, and the ability to link any location to any other. Nobody has yet come up with an integrated tool that makes all of that as easy to do for content and functionality as the spreadsheet did for numbers and calculations, but some are coming close. Trellix not only includes a wide range of functionality of its own, it also incorporates third-party services that can be easily embedded into its website framework, such as Miva webstores or TimeTrade online scheduling. FrontPage is supported by a growing number of third-party plug-in services that can be added to a website from a FrontPage menu the latest example is PayPal.
All of these developments are bringing the killer app closer the Web as a universal spreadsheet for the assembly of online content and automated services that meet the needs of business, society and individuals. The killer architecture that enables it already exists, and it's called web services. All we need to do now is fill in the gaps with an application framework that will make it accessible to ordinary users. With its announcement today, Trellix has built a significant bridgehead across one of those gaps.
posted by Phil Wainewright 3:00 AM (GMT) | comments | link
Wednesday, August 07, 2002
Is WS-I FUBAR?
With a plethora of web services standards initiatives under way in dozens of separate working groups at an assortment of different industry bodies who will co-ordinate all this activity to ensure a consistent result, wonders Eric Knorr: "The tower of proposed standards for Web services authentication, business processes, and security is already teetering and could topple."
One would have thought that would have been exactly what the Web Services Interoperability organization (WS-I) was set up to achieve (co-ordination, that is, not toppling). But how can it when it is an undemocratic body set up with the explicit aim of excluding minorities (ie Sun Microsystems)? "Under the organization's current bylaws, an unelected cadre of players sets the agenda," he writes, and goes on to quote an insider who doubts whether WS-I could ever have credibility as an impartial venue for reconciling standards interoperability conflicts.
I've been looking through my previous comments on WS-I and have been surprised to discover how much scorn and ridicule the organization has wrought from such a mild-tempered individual as myself. There is definitely a serious image problem developing here:
Tuesday, July 09, 2002 "The shenanigans over Sun's (non-)participation ... have become so repugnant that it reminds me of the doublespeak in George Orwell's 1984 a world where Truth means Propaganda, Peace means War, and Interoperability means Proprietary Lock-in."
Friday, June 28, 2002 "WSDL rhymes with 'whistle', and WSCI sounds like 'whisky' — although I've not heard the Web Services Interoperability consortium (WS-I) being called 'whizzy'."
Thursday, May 16, 2002 "Microsoft's choice of the internal codename 'foo' for the WS-I ahead of its launch is intriguing ... 'foo' ... is generally traced to the WW II era Army slang acronym FUBAR (`F***ed Up Beyond All Repair'), later modified to foobar. Is this perhaps a clue to Microsoft's true intentions with regard to WS-I?"
Former British PM Margaret Thatcher once famously remarked, "You can't buck the market." She was referring to the financial markets, but she might just as well have been talking about any shared network system. When you participate in a network, you implicitly agree to abide by its rules. Any attempt to selectively opt out of certain rules in favor of your own proprietary mode of operation is doomed to failure (except in two very special circumstances: A, the network is crumbling anyway, or B, the network ends up adopting your rules).
Two consecutive entries in Timothy Appnel's blog happen to give some useful illustrations of this point (it's probably not happenstance, actually, since Timothy writes a lot about network effects).
The first brings the news that Nextel (finally) Gets Text Messaging Religion. Amazingly, this US mobile phone carrier believed its subscribers would find text messaging useful even though it refused to interoperate with other carrier's text messaging services. (As an aside, I must say it beggars belief how long it is taking people in the telephone network operators business to understand how networks operate).
The next posting quotes BEA's CTO Scott Dietzen writing in Web Services Journal about the two supposedly rival web services architectures, Microsoft's .Net and everyone else's J2EE:
"The World Wide Web is simply growing too quickly and too diversely (with phones, handhelds, on-board navigation systems, entertainment/gaming systems, etc.) for it to be in the financial interests of any vendor to break interoperability."
Nextel's failed attempt to confine users to its own private enclosure of text messaging is a warning to all vendors of the dangers of spurning interoperability. This is a virtuous circle, too. Each new breakthrough in interoperability increases the strength and universality of the network, further diminishing the power of any vendor (even the most powerful) to resist.
Perhaps the sinking in of this message is what is now fuelling the share price falls and management turmoil at the major entertainment and media conglomerates, including AOL Time Warner, Vivendi, Sony and even Disney. They expected one day to control the Internet, but instead they are discovering that it is the Internet that will dictate the terms on which they are going to have to operate. It is the network the demosphere that is in charge.
posted by Phil Wainewright 2:30 AM (GMT) | comments | link
Monday, August 05, 2002
Ray Ozzie goes public
The founder of Groove Networks (and before that, creator of Lotus Notes) has forsaken the safety of his company's shared private spaces to begin publishing a weblog. There's not much there yet, but its very existence has already excited comment from Jon Udell, which was picked up by Dave Winer, thus generating plenty of traffic.
Why so much attention? Quite simply, because Ray Ozzie is more or less the patron saint of corporate online communication. His career to date has been entirely devoted to enabling organizations to share knowledge, information and gossip within enclosed, private spaces. And now he's venturing into a public space or proposing to, at least, for as he states (and as Jon Udell rapidly elaborated on): "Forcing myself to partition internal vs. external on a daily basis would truly be a mindset change..."
But as Jon subsequently pointed out, the very fact of being online in a Radio weblog is already exposing internal information (no doubt unwittingly), an example of one of those unintended consequences (this phrase is rapidly growing on me, by the way) that are a daily feature of the networked world.
This debate about internal vs external, private vs public, reminded me of a recent posting to the klogs discussion group, about whether 'knowledge weblogs' (ie klogs) are relevant in academic circles. One contributor, while welcoming the contribution that weblogs could make, cited some possible objections, the last of which was this: "Fear that someone else will pick up their ideas and work them out before they do."
I think this highlights one important aspect of a highly networked world that we are all going to have to get used to. There are very few genuinely original ideas in the world. Someone, somewhere has inevitably already come up with the same idea. By pooling your thoughts with theirs, both of you will likely progress them further than you could have done individually (or maybe someone else watching the exchange will have a new insight that takes the idea further than the pair of you). The more open the network, the more everyone can feed off each other's ideas. The less open it is, the more slowly everyone progresses.
So which is better? I think the answer is that, in an extensive open network, the one thing you can be sure of is that someone else already has the same idea as you. If you deny that fact, you relegate yourself to coming in behind them. If you accept it and embrace the network, you have a chance of participating in their success. (I have a feeling this has been said better by someone at Microsoft, but I can't recall the reference just now. Perhaps someone reading this will be able to refresh my memory).
There is still, of course, a place for private shared spaces. Planning especially of execution still has to be done within the enterprise. But innovation and speculative thought is something that benefits from the widest possible network, and therefore must always be exposed for best results. This is counter to the way things have customarily been done in recent years, and therefore is going to be difficult for us to learn (and in the learnng process, mistakes will be made). But it is the new way things will be done in the web-connected future.
posted by Phil Wainewright 12:41 PM (GMT) | comments | link
Assembling on-demand services to automate business, commerce, and the sharing of knowledge