Should you start building out a service-oriented infrastructure right now, or is it better to wait till you have enough services in place to justify the investment?

	express delivery
	• print  • comment
	WebMethods has assembled an infrastructure offering that is designed to offer its customers a roadmap into SOA: Its integration platform service-enables existing resources Its newly-acquired Fabric tool adds service assembly and management An SOA roadmap leads on to process orchestration and activity monitoring The aim is to future-proof the company's offering But it faces stiff competition in the SOA market
	Glossary terms: SOA, EAI, BPM, orchestration, loose coupling, lookup tool Sponsored links: your text link here

You might as well ask whether the chicken or the egg should come first. If an organization is going to start assembling services, it's going to need a service-oriented architecture (SOA) in place. But if it goes ahead and builds the infrastructure first, it might yet discover there are precious few web services out there to manage.

WebMethods believes it has an answer to this quandary. Over the past six months, the integration software vendor has been furiously assembling the components necessary to back up its freshly minted claim to be "the world's first [true] web services infrastructure company". With this infrastructure, so the argument goes, its customers will use the webMethods Integration Platform to deliver services; webMethods Fabric to marshal, manage and monitor them; and webMethods Portal to view them.

Linking its existing integration platform to the recently acquired Fabric — the first fruits of which are due this month in a feature pack upgrade — means clients are no longer gambling on adoption of a web services-based SOA becoming the prevailing trend, says the company. Simply by adopting webMethods as their integration infrastructure, they can guarantee — if only internally — that the pieces of the jigsaw will all be in place.

"Initially there was a misunderstanding that we were throwing the Integration Platform aside and replacing it with the Fabric," says Jim Ivers, VP or product marketing at webMethods. Instead, Fabric's role is to add a new layer of service assembly and management, which will extend the existing product's application integration capabilities into the more loosely coupled SOA environment. "There will be places where we'll take functions from the Platform and we'll Fabric-enable those functions, but the Integration Platform as it exists not only won't go away, I see it as having some good characteristics for the Fabric," he explains.

"Any [assembly] tool, even as good as the Fabric, is only as good as the web services building blocks that are available to build new applications within it. What the Integration Platform does is allow customers to reuse their existing assets, expose them as web services and build this really rich population of web services," he concludes.

Ian Charlesworth, senior research analyst with the Butler Group, endorses this emphasis on service-enabling existing applications so they can link into the SOA: "There is no 'Web services fairy' that will materialize in the still of the night and sprinkle magical SOAP-dust over the relevant applications. The fact of the matter is that while the process of application development remains only a relatively small part of the application lifecycle, it forms the basis for the successful deployment, integration, and ongoing management."

$32 million gamble

In effect, webMethods has gambled on web services infrastructure so that its customers don't have to. The most visible element of this strategy has been its acquisitions of The Mind Electric (for Fabric), Dante Software (for business activity monitoring) and Netegrity's portal product, formerly known as DataChannel. The purchases will cost webMethods around $32 million, which — despite its difficult financial position following a series of losses — is a relatively small sum to pay when it has nearly $200m in cash reserves — especially if it does give it all the pieces it needs to satisfy its customers' needs.

WebMethods is convinced the gamble will pay off. "Web services was taking our customers in a completely new direction, which was adopting an SOA as something that was feasible and do-able within their company," says Ivers. That was over and above what webMethods had previously been offering: "Integration is making existing applications work together better but SOA is about building new applications and business functions," he explains.

But by committing so firmly to TME's web services-based architecture, webMethods risks having jumped too soon. There are other ways of achieving SOA, says Steve Craggs, independent analyst with Saint Consulting and vice-chair of the EAI Industry Consortium: "This year we will see whether web services is here to stay or whether it is just another passing fad. Some of the vendors may be waiting to see what happens. webMethods has bravely staked its colours to the mast. Others will carry on trying to sell into the EAI market and of course they will support web services. But if they tie their colours to the web services mast and web services falls over, they are going to look a bit silly."

Services roadmap

For webMethods, what matters is giving customers a roadmap and the ability to pursue it as soon as they're ready. "Everything my customers do today for pragmatic integration is immediately available to the Fabric," says Ivers. "So we're helping them solve the pragmatic problem, but we're also propelling them into an SOA very quickly. A lot of the companies that compete against us don't have that integration platform."

Acquiring Fabric has advanced the company's understanding of services architectures, Ivers admits. "TME [The Mind Electric] has taught us a lot about SOA and the way we need to go forward with that; and we're looking at the way we can take our BPM products and make them more savvy as far as orchestrating web services goes.

"What we did is really take a hard look at where we wanted to go as a company. Integration is something that continues to be in demand and in fact it's growing. But the direction in which we were going was in helping our customers build an SOA and leverage their systems assets in that architecture. And then we were taking that as far as building new processes and taking what they had and use them in business processes. We were seeing an increasing emphasis on BPM in those companies. And then finally, once you tie in all those things, it's about analysing methods around those business processes."

While Fabric is an add-on at present, future releases will aim to make the boundaries between the products increasingly invisible, says Ivers. According to Guljit Khurana, VP of product technology and a former TME employee who has been working on the integration between the two products, the benefits are already visible to users: "On the integration platform you can get a real-time view of all the services that are available, and then you can expose any services as a web service. The [user interface] will allow webMethods users to drag-and-drop Fabric services directly into the integration platform and vice versa, so it's really a rich two-way integration."

Future payoff

The gamble should pay off if webMethods can convince customers that it has future-proofed its offering. "There's a lot of talk about integration becoming a commodity and so on," says Ivers, "so this puts it in people's minds that webMethods is taking the next step forward."

That should certainly help to retain the loyalty of existing customers. But webMethods has to tempt new customers to try out its platform if it is to reap a bigger return for its efforts. "Gartner says that 60 percent of all application development by 2006 is going to be handled in SOA," Ivers points out.

As a result of its acquisitions, webMethods can now compete for a slice of that 60 percent. But it faces stiff competition, not only from its traditional EAI rivals, but also from server platform vendors including IBM, Microsoft and BEA; from application vendors who are moving into service-oriented integration, such as SAP and Siebel; and from SOA specialists like Systinet and Cape Clear. To stand out from the crowd, webMethods needs to coax at least some of its customers to begin following its SOA roadmap, so that it can establish an early track record in delivering proven implementations of its service-oriented infrastructure.

More on this topic