In-Depth

Microsoft Does Have a SOA Strategy

Redmond was conspicuously absent as the Service-Oriented Architecture hype exploded -- turns out it had a plan after all.

• By Ed Scannell and Chris Kanaracus
• 08/01/2007

Microsoft's ruthless campaign against open source software and, with somewhat less vengeance, its reluctance to join the Software as a Service (SaaS) movement are the latest examples. Only after years of bloody public jousting did Microsoft finally seek ways to peacefully coexist with the open source world through deals like the one signed with Novell Inc. last November. And it did acknowledge the market presence of SaaS when it introduced the Live versions of Windows and Office, both of which are still incubating.

SOA Takes the Stage
Enter Service-Oriented Architecture (SOA), which for several years has been developing into a transformative force in enterprise IT. SOA, as a concept, has been embraced by Microsoft's usual cadre of competitors, most notably IBM Corp. and Oracle Corp. It has made its way onto the radar screens of the more forward-thinking IT shops, and established standards groups are shaping the rules by which to play.

But true to character, Microsoft again appears to be crafting its own rules and vision. The company has so far declined to participate in certain key emerging industry standards relevant to SOA. It has a different perspective on what SOA is and a different approach for crystallizing its vision. Microsoft has even shown a certain reluctance toward using the term itself, choosing instead to talk about its "services-oriented" or "service-enabled" approach.

The more vocal critics claim Microsoft's approach to SOA not only goes against the technical grain of competitors, but may also not be in the best interests of customers. They believe the company's approach is too tied to pushing sales of its core desktop and server products, which are more expensive, complex and proprietary than alternative offerings.

"Microsoft is primarily concerned with its [own] business strategy. It wants to continue to produce these fantastic profits but that runs counter to what many IT shops are focused on, which is cost-reduction, simplification, consolidation and modernization," says Dana Gardner, principal analyst with Interarbor Solutions LLC in Gilford, N.H.

Gardner and other industry observers charge that Microsoft's business model too strictly requires users to buy into its bread-and-butter operating systems, applications, run-time environment and tools before they can start piecing together a customized SOA that best serves their needs.

But top Microsoft executives disagree with such assessments. They say all the basic building blocks of the company's services-oriented plan can be mixed and matched with a wide range of competing technologies, that the resulting SOA-enabled applications will offer the agility needed to flourish in the fast-paced Web 2.0 world, and that they can do so cost effectively compared to competitors.

"It was clear that if the services-oriented world was to coexist with the real world, we needed to move data efficiently around the organization in a way that was transport-agnostic. This is why we invested in Windows Communications Foundation [WCF]," says Steve Martin, director of product management for Microsoft's Connected Systems Division. WCF is Microsoft's Web services stack shipped as part of the .NET Framework 3.0.

Besides the ability to move data around easily, Microsoft is placing an emphasis on the agility of the SOA-enabled applications. The reason, according to Martin, is that most custom applications live between six and 15 years in an organization, while service-oriented applications live three to six months.

"This is what you're investing in when you take a services-oriented approach -- the ability to rapidly evolve an application because you need to change things in near-real time," Martin says.

'Good, Better and Best'
Microsoft is not selling a big, fat SOA stack, Martin argues. Instead, the company is taking its "traditional" technical approach to the SOA market -- one that is focused on empowering individuals, he says.

IT shops can simply use WCF to service-enable their existing systems, he says. Or they could go further and use Microsoft's best selling Visual Studio suite of tools to build new services. A third and optional step, he says, would be for users to then combine these assets with BizTalk Server, which allows them to wrap legacy technologies with the new class of services. Giving users and developers these options to start small and flexibly build capabilities as their business requires them is Microsoft's "good, better and best" approach to the SOA market.

"People should be able to fly at the altitude that is most appropriate for them," Martin says.

Another central technical element to Microsoft's services-oriented strategy is virtualization. David Greschler, director of virtualization strategy at Microsoft, says virtualizing applications may be the best way to SOA-enable many of Microsoft's best-selling applications.

"It's important in an SOA-enabled world that we think about how to move these applications into that world," he told Redmond in an interview after his keynote address at the SOAWorld 2007 show in late June. "One way to do that is to virtualize them. By tying those apps to policies, by en-abling them to be delivered in real time to whatever device needs them, we've taken the same characteristics that one would see with a Web app and made most any Windows app Web-enabled. It can be part of an SOA-like strategy."

But for all of Martin's talk about the company's ability to provide SOA a la carte, Microsoft is also taking some stabs at a market IBM and others have long ruled: the world of high-performance back-end megasystems that power verticals like the financial services industry.

In June, the company released .NET StockTrader, a trading application based on ASP.NET and WCF. Microsoft highlighted its ability to interoperate with IBM's WebSphere Trade 6.1 sample app and released benchmarking results that showed significant performance improvements over Trade 6.1.

Forrester Research Inc. analyst John Rymer says it's clear that Microsoft wants to show that it's a viable alternative to products such as WebSphere -- and not a supporting player. "Obviously, they're trying to sell it. The problem is IBM is seen as enterprise and they're not," he says.

Rymer says the .NET StockTrader benchmark tests were well-documented, but like any benchmarking, the results should not be swallowed whole. That said, Microsoft is now well-positioned to become a stronger player in enterprise systems, according to Rymer.

Microsoft is backing up its technical investments in this market by bringing on high-powered talent, and from its archrival no less. Earlier this year, it scooped up former IBM Fellow and Chief Architect Donald Ferguson, who's generally credited with turning WebSphere into the success it is today. While his role at Microsoft has not been made clear by the company, Ferguson -- someone long-steeped in the SOA tea -- figures to have a hand in shaping the company's services-oriented strategies in a way that could exploit any potential IBM weakness.

Fatal Flaws?
On the surface, then, Microsoft does seem to have a straightforward and fairly substantive SOA strategy, one Martin details in depth for Redmond (see "Microsoft and Service-Oriented Technology"). But industry observers stress that there are clear limits to it.

Forrester analyst Randy Heffner says that as a company, Microsoft's philosophy is to be "standards-based at the edge. But when you talk about standards that affect internally how things go, then Microsoft doesn't play in that ecosystem nearly as much."

An example of this is Microsoft's decision not to join a phalanx of other vendors -- including IBM -- in endorsing a pair of potentially key standards for SOA: the Service Component Architecture (SCA) and Service Data Object (SDO) specifications. Both are still in development under the watch of the OASIS standards body.

Such standards are too closely tied to rival technologies and platforms for Microsoft's taste, Heffner says: "That would be a hard pill to swallow. Microsoft doesn't want to do the tools that will help people use some other platforms."

Asked directly why Microsoft hasn't joined SCA and SDO, Martin somewhat deflects the question, saying, "SCA is a great endorsement of the work we have done on the Windows communications side. It tells us that others think the strategy of building a set of technologies that are transport payload-agnostic was the right way to go."

While Martin and his colleagues at Microsoft may have crafted an array of SOA-enabling technologies, it's an open question how many enterprise customers are prepared to immediately embrace them. Eric Manley, a consultant who's working on SOA-related initiatives for a Fortune 500 company, says his company is still preoccupied with purely human concerns.

"We're getting there. We've got a lot of point-to-point implementations. I think the bigger challenge is getting over more of the governance, who decides what," he says.

Manley says his company currently has groups working to determine exactly what a runtime environment should look like. Once that's determined, he can start to define the minimal requirements in order to get into that environment, and what the service-level agreements should be.

"Each project is kind of funded from its own money, so [different departments] are very apprehensive about building something that somebody else could consume. You start out with the deck stacked against you," he says.

Manley says that so far, there are only scattered instances where SOA-related efforts have resulted in reusable services.

Like many IT shops, Manley's company is several years behind Microsoft in terms of the underlying technology being used. He says only now is his company moving from the 1.1 version of the .NET Framework to 2.0. He adds that his current environment also includes a healthy amount of IBM's WebSphere technology, although the client machines are predominantly Windows boxes. His biggest challenge, however, is not technology-related but centers more on organizational and social changes.

"Whether it's .NET, Java, Oracle, SAP, who cares? It's working through the governance and who pays for and decides what that is-really, to me in my world [that's] the real challenge," he says.

Some believe the inherent flexibility of SOA-based software, in addition to the perceived greater financial incentive to diversify to multiple platforms, makes more sense for companies like IBM and Oracle than for Microsoft.

"With SOA [Oracle and SAP] can take existing services and components from one set of business apps over to a ripe vertical market and reuse 70 or 80 percent of them and then tailor the remaining 20 or 30 percent to that specific industry. Suddenly, SOA makes a whole lot of sense for ISVs," Gardner says.

Survival of the Fittest
Now that Microsoft's services-oriented strategy is finally crystallizing, there are two ways to think about the future: Is Redmond simply too late to become a major SOA player? Or will the company emerge fairly unscathed from the sector's initial growing pains, poised to become a viable alternative?

Heffner, for one, is not betting his entire bankroll. "They will survive," he says. "Will they get as much penetration in the enterprise as they want? ... I don't [see them becoming] Java killers any more than they've been in the past."

But it may not be necessary to for Microsoft to go lion hunting when it can go trawling for fish instead. Microsoft has a history of going after high-volume, low margin markets, and with SOA still a high-end, high-margin business, the company may simply be sitting and waiting for the commoditized market to materialize with the arrival of low-cost SOA-based tools and platforms.

But whatever the hunting weapon Microsoft chooses for this expedition, there's a good chance the route it takes won't be on a map.