Building applications that are web services-ready sounds like good advice, but how do you do that when the web services don't exist yet? Writing on The need for a dynamic invocation framework on WebServices.org, Harshal Deo outlines a method of writing applications that will be able to switch from calling a local native API to calling one of several alternative web services without any need for reprogramming:
"Some runtime parameter signals to the application that a specific task is now carried out by a Web Services rather than a native method, and sets up that task accordingly. What about if the Web service changes? In an ideal world, if the WSDL file for a Web Service changes, there should be no need to recompile the application, or change the code."
Harshal Deo works for Asera, which is a pedigree that makes his recommendations well worth heeding. The company has been building applications using a service-oriented architecture since before anyone started talking about web services. That adds up to a lot of practical experience of building solutions that stand up in the real world.
posted by Phil Wainewright 1:27 PM (GMT) | comments | link
Putting the whizz into web services
Those in the know pronounce certain web services standards as words rather than spelling out the letters in the acronym. Hence WSDL rhymes with 'whistle', and WSCI sounds like 'whisky' although I've not heard the Web Services Interoperability consortium (WS-I) being called 'whizzy'.
No doubt there are some teams working hard on standards that will mop up the few remaining words in the English language beginning with 'whis'. My dictionary tells me to expect WSPR, WSCR and WS-T, with WS-Bang making a raucous finale.
Once these alternative pronounciations catch on more widely, we'll probably start to see them cropping up in headlines, though the most obvious ones will only make sense in the context of bad news or setbacks "WSCI on the rocks", "WSDLing in the dark" and, as in my posting yesterday, "WSCI sours". But some will have a more positive gloss. As Lauren Bacall memorably once said ... (see next posting).
posted by Phil Wainewright 1:21 PM (GMT) | comments | link
Thursday, June 27, 2002
WSCI sours for ebXML as BPML opens 1.0
BPMI.org yesterday published BPML 1.0, the first public draft for release 1.0 of the Business Process Modeling Language. At the same time, it strengthened its industry clout with the creation of an alliance with the Workflow Management Coalition (WfMC), which represents workflow vendors.
According to BPMi.org chairman Ishmael Ghalimi who is also chief strategy officer of Intalio, the BPM startup that has been the prime mover behind both BPML and WSCI the latest version of BPML is "a superset" not only of the previous release BPML 0.4, but also of IBM's WSFL specification and of Microsoft's XLANG technology.
The combination of BPML 1.0 and WSCI brings interoperability between various flavors of BPM, Ghalimi told delegates at Information Age's Agile Business conference in London today. "You can use WSCI as an interoperability protocol between multiple business process models," he explained.
BPML 1.0 is open for discussion prior to the release of a final draft, expected in August. It will be royalty-free, along with WSCI and all other BPMi.org specifications, Ghalimi pointed out. The group plans to introduce a specification for BPM notation in Q3 this year which the WfMC has agreed to endorse and a query language specification is due early next year.
Not everyone is happy about the announcements, however. Adding fuel to speculation that there is no love lost between BPML's backers and supporters of the UN-sponsored ebXML, Jean-Jacques Dubray, chief architect at eXcelon and project leader of the OAG's eCommerce initiative, slammed WSCI yesterday in a personal statement: "This is merely a non-refereed paper which look at some deficiencies of the web services specification framework, managed by the W3C, and propose some solution. By contrast, ebXML is a stable set of specifications complementing each other and driven by a global architecture specification. Nobody in ebXML develops 'rogue' spec, which main purpose it to make a marketing 'coup'."
I've oftenwondered what the A in EBITDA stands for. Now I think I've got it: Earnings Before Interest, Taxes, Deception and Accounting irregularities.
posted by Phil Wainewright 12:12 AM (GMT) | comments | link
Tuesday, June 25, 2002
Gaps for startups in web services management
Small firms will be the main drivers of innovation in web services, even though all the media focus is on what big names like Microsoft, Sun and IBM are up to. While most of the mainstream analysts are saying it's the big names that will dominate, everyone should take that with a pinch of salt. They're saying it because it's the safe thing to say, which no-one will criticize if they turn out to be wrong, whereas endorsing startups is a highly risky strategy.
Fortunately not all analysts are cast in the same mould, and the team at Summit Strategies are more willing than most to put their necks on the line. A new report by Summit analyst Dwight Davies highlights some of the ways in which smaller vendors are making a contribution, as he explains in this interview with Eric Parizo of searchWebServices:
"Aggregation and routing companies ... provide the virtual glue to connect Web services once they are built ..."
"Integration and process flow vendors ... specialize in smoothing out the connection between business processes and Web services ..."
"Management companies ... handle anything from overseeing the development of a single Web service to taking on responsibility for the performance of an entire architecture ..."
"There are all these adjunct areas management, quality assurance, security that you need to address if you are serious about building enterprise web services solutions," he sums up. The article itself names six of the nine startups profiled in the report.
Another of Eric's articles last week focussed on web services management, prompted by an interview with highly rated startup AmberPoint and one of its customers. "A number of companies ... are working on or have released software products as of late to help companies monitor, fix or improve their web services," he writes. I suspect there will be a whole lot more of them before things start to settle down. Even with so many entrants introducing new tools and technologies, there are still several aspects of web services management that have barely been addressed as yet.
Bowstreet's founder Frank Moss explained why this is at a meeting in London recently. Just as client-server was harder to manage than the mainframe, so web services will be harder to manage than client-server, he said. The simple reason is that there are more things going on that need to be managed. That's where the best opportunities will lie for startup companies, and when mainstream vendors (and analysts) finally wake up, those startups will become prime candidates for acquisition. Of course that's exactly what happened to Frank's former company Tivoli, which provided management tools for client-server environment, when it was acquired by IBM for $1 billion in 1995.
posted by Phil Wainewright 6:27 AM (GMT) | comments | link
Monday, June 24, 2002
Platforms by stealth: Macromedia, Progress
How can an up-and-coming vendor win market share for its platform against the big guys? The trick is to sell the product as an adjunct to something else rather than as a platform in its own right. This "platform by stealth" strategy was outlined by Cape Clear Software co-founder David Clarke at a recent meeting: "It's our belief that you become a platform by having a good tools strategy," he explained. "Over time, your runtime becomes a platform, and ultimately [it becomes] THE platform."
Two vendors that seem to have excellent "platform by stealth" strategies in the web services sector are Macromedia and Progress Software. Both are established companies with a substantial global customer base, but nobody would rank either of them on a par with the likes of Microsoft, BEA or Oracle. Yet all three of those giants were once rank outsiders themselves, before a shift in the computing landscape opened up the opportunity for them to establish their product as a leading platform.
Macromedia's strategy in particular I find breathtaking in its scope. With April's launch of its MX family, it put in place all of the elements required to become the presentation platform of choice for the web services era. Look at what it has going for it:
The client piece of its platform, Flash MX, is already accepted among Internet users, to the extent that Macromedia claims an incredible 98% already have a Flash player installed. The fact that Flash got on those desktops as a means of displaying frivolous eye-candy doesn't matter any more. Now that it is there, it becomes the most convenient way of presenting complex application choices and real-time information feeds to Web users.
The server piece is ColdFusion MX, based on one of the most popular application server platforms at the volume end of the market, but rearchitected in this release so that it will run on enteprise-class J2EE servers from the likes of BEA, IBM and Sun. It's cheap, easy for developers to learn and use, and now scalable all the way from low end to high end.
The tools piece is Studio MX, the centerpiece of which is Macromedia's web design tool Dreamweaver, already a top choice among professional website designers. They already know how to use it and they like using it. Hooked into all the capabilities of the rest of the MX family, it makes it easy for them to move beyond building web pages and into building distributed, component-based web services applications using Macromedia's technology as the platform.
Progress Software isn't as far along the curve as Macromedia, which has already shipped its MX family, but its service oriented architecture (SOA) strategy, announced at the beginning of this month, contains some compelling ingredients:
Sonic Software, which Progress span out as a wholly-owned autonomous subsidiary last year, contributes SonicXQ, which packages Sonic's Java messaging technology together with routing, transformation and adapters to create what it calls an "enterprise service bus".
The core Progress platform is an integrated application server, database and development environment, used by a 2000-strong ISV partner base to create business applications. Later this year, a web services toolkit will allow those partners to migrate to an SOA strategy by exposing the individual business functions from within their applications as web services. Progress is also actively recruiting partners from rival vendors whose channel strategies have been less robust of late, including Oracle, Novell and Informix
The company's market presence is masked by its partners' private-labelling of its technology, often to the extent that customers themselves don't realize they're using Progress in several different divisions and applications. Adding SonicXQ's messaging while migrating the legacy base to a service-oriented architecture will start to make those disparate components look more like an enterprise-wide platform all of them based on Progress technology.