There is no longer any question about it: Companies that want their application infrastructure to keep pace with fast- changing business needs are deploying service-oriented architecture (SOA). But have they the time and the expertise to build an SOA platform themselves, or are there off-the-shelf products available that will provide all or some of their SOA needs?
For most companies the answer to the first part of this question is simple. Deploying SOA is a complex challenge beyond the means of all but the most sophisticated IT users, and even those users have such an urgent need for the benefits SOA can bring that sourcing a platform from one or more vendors off-the-shelf offers a quicker and cheaper means of meeting their needs.
The second half of the SOA platform conundrum is not so simple. Sensing a boom market in the making, established enterprise software vendors are dressing established products in new SOA clothing, and new players are touting innovative ‘pure play’ SOA products. Each boasts functional strengths that might appeal to buyers set on a best-of-breed procurement strategy – so long as they can interoperate with the components of their rivals. Others may even offer a ‘total solution’ – but will it be adaptable enough to meet every customer’s needs now, and in the future?
In this Information Age guide to SOA platform vendors, we offer a strategic overview of five of the biggest contenders – but they are far from being the only ones to choose from.
IBM's integrated vision
FEW companies have wrestled so hard for so long with the challenges of systems integration as IBM. During the 1980s and 1990s, IBM’s entire software strategy was based on a succession of strategic initiatives aimed at overcoming the architectural compatibilities within its own software products. More recently, the company has caught the ‘open’ bug, and made standards-based development a central tenet of its product strategy.
Given this history, it should be no surprise that according to analysts such as Ovum’s Ian Charlesworth, “IBM now has a breadth of [SOA product] offerings that outreaches any other vendor.”
Indeed, the products listed in what might be called IBM’s SOA portfolio constitute an embarrassment of riches. At the programming environment level, the company’s Rational Software range offers a broad set of programming tools that conform to J2EE and Eclipse standards, and that also now routinely support the business process engineering language (BPEL). At the SOA communications layer, alongside IBM’s core WebSphere Message Broker there are mature alternatives such as WebSphere MQ Series. While at the management, governance and business process management layers, IBM offers a raft of WebSphere-branded products that support or manage business modelling, process serving, orchestration, provisioning, and asset and business rule registry and repositories.
While IBM has the capability to provide a one-stop-shop for SOA infrastructure that is not its stated strategy. Rather than setting out to provide a universal SOA solution, “our strategy is based fundamentally around architecture,”says Garry Gomersall, IBM director SOA champion and sales leader. Specifically, IBM has built an SOA reference architecture designed to allow IBM products to be “plug-compatible with those of any other vendor” and so provide users with the widest possible set of SOA deployment options.
It is arguable that this architectural approach differentiates IBM from those vendors which assume that customers will make a strategic investment in their
specific middleware products. However, IBM’s critics can counter that product licensing is no longer where IBM expects to reap the richest rewards. Instead, IBM’s underlying SOA market strategy is to invite its customers to enter an ongoing services-based relationship.
The full extent of IBM’s SOA service ambitions are clearly reflected in its acquisition earlier this year of Webify. Previously a close business partner of IBM, Webify provides off-the-shelf SOA frameworks for vertical markets. Currently these frameworks include a set of SOA-based assets that enable healthcare companies and insurers to accelerate the delivery of regulatory compliant systems.
IBM is currently integrating Webify and its products into its WebSphere portfolio, where it will be encouraged to generate more bundled offerings aimed at a variety of vertical sectors, all of which will be expected to become lucrative sources for IBM’s Global Services business.
SAP weaves its services tale
SOA is both a threat and an opportunity for enterprise application vendors like SAP. The threat is that, one day, SOA will make it easier and cheaper to buy software in small, granular chunks – services – rather than as big, expensive and inflexible, monolithic applications, such as MySAP.
On the other hand, businesses will still need a lot of application functions, and if these are delivered as services, they will need to be managed, orchestrated, and carefully integrated. Who better to provide these services, and manage them, than the big, trusted and incumbent supplier of enterprise-class applications?
SAP has not always managed to convince the most sceptical analysts that it can or should do this, but its position is clear nevertheless: It wants to be at the heart of SOA – as a services provider and as a platform provider. From their earliest days, Hasso Plattner, SAP’s co-founder and technology visionary, and later Shai Agassi, now president of the product and technology group at SAP, have pronounced componentisation and service orientation to be a good thing – regardless of any apparent threat to SAP’s traditional “big app”, monolithic approach.
In the past three years, that commitment has translated into a comprehensive strategy, backed with a large investment in R&D, a series of products, and a detailed SOA roadmap. If there is any doubt about SAP’s SOA strategy, most of it relates to the willingness of its customers, partners, developers and even its own commercial managers to fully support such a dramatic transition.
The strategy has four elements. First, there is NetWeaver: the integration broker, application server and services orchestration engine, that is at the heart of the MySAP Business Suite, handling all interactions between the applications (among them SAP’s ERP, supply chain management, and CRM products).
Second, SAP has adopted web services standards and methodologies for developing and integrating applications, and a loosely coupled and distributed architecture for linking these together – its Enterprise Services Architecture (ESA).
Third, it has introduced xApps – composite applications, based largely on underlying MySAP functions and other web services that can be built using NetWeaver and other tools.
Finally, SAP has committed to open up its core ERP, CRM and supply chain suites, converting these tightly integrated, proprietary application suites into accessible collections of integrated web services.
So far, says Martin Tenk, head of technology for SAP in the UK and Ireland, the company has repackaged about 300 application functions as “enterprise services”, with similar numbers covering the other packages. By the middle of next year, when SAP will introduce the next generation of its MySAP business suite, it should have delivered between 1,000 and 1,200 services.
But questions remain: Analysts say SAP lacks many of the orchestration, management and governance tools that SOA demands – Gartner thinks it will need to make acquisitions.
Rivals point out that NetWeaver’s acceptance in the market is unproven, because it is included “for free” as part of MySAP business suite; there are few users outside the SAP-using community. And while SAP has delivered more than 500 enterprise services, most users and partners have done little to exploit the new openness.
Two to three years since it launched the concept, SAP admits there are no more than a couple of dozen xApps.
But it is early days for SAP – and its customers. “Most customers are just dabbling with web services. But they will eventually deploy SOA,” says Tenk.
When that happens, SAP is likely to be chosen, at least, by a good number of its own 33,000 enterprise-class customers.
Microsoft in SOA stealth mode
Of all the major enterprise software players Microsoft has probably done least to explicitly position itself as an “SOA platform” provider. Neither the company’s customers nor its rivals should interpret this as evidence of either a lack of commitment to service-oriented software delivery, or any lack of capability.
Steve Ballmer publicly committed Microsoft to a services oriented future as long ago as 1999. In a speech to industry partners and developers Ballmer (then Microsoft’s president, now its CEO) declared that a fundamental shift in programming practice was imminent. The age of monolithic applications was drawing to a close, and in future organisations would dynamically create Internet-based business processes from disparate and distributed software assets.
Just 10 months later, Microsoft launched the .NET programming environment, and since then it has diligently tracked and participated in a variety of standards-based initiatives focused on web services that has produced an impressive and comprehensive set of service-oriented infrastructure products – with plenty more products waiting in the pipeline.
Today, Microsoft can fairly claim to have all the major SOA bases covered. .NET’s credentials as an enterprise-class programming environment are unquestionable, and although Microsoft’s portfolio contains nothing as explicit as an enterprise services bus (ESB) the rapidly evolving Windows Communication Framework (WCF) already provides a rich, if still largely Microsoft-centric set of integration programming tools.
In the realm of service management and governance, the Microsoft Operations Framework offers a comprehensive ITIL-based platform, supported at an asset management level by the now mature Microsoft Operations Manager (MOM). At the business process management level, Microsoft’s Biztalk is enjoying growing market popularity, and BPM functionality is becoming a default component of its CRM, ERP and related business application suites.
Critics of Microsoft’s strategy point to the lack of any explicit SOA platform product, and some also persist in the view that Microsoft's portfolio is geared primarily to winning the hearts and minds of developers rather than of business decision makers. As far as the former criticism goes, Matt Deacon, chief architectural advisor of Microsoft’s developer and platform group, argues that the critics are missing the point. “SOA is an architecture, an approach to solving [business] problems,” he says. So, although “we too could rebrand Biztalk as an ESB,” he says, instead Microsoft has opted to take an architectural rather than a product-based approach to delivering SOA functionality that is ultimately more open and less restrictive than other proprietary approaches.
As for the notion that Microsoft is still offering essentially a programmers view of SOA, that will start to change soon too, says Deacon. In particular, he says “Longhorn will be a pivot point” for Microsoft’s relationship with corporate IT architects.
The next iteration of the Windows Server platform will be based on SML, the service modelling language standard recently endorsed by Microsoft, BEA, BMC, HP, IBM, Intel and Sun, among others. Microsoft plans to make it a
central unifying element in its server infrastructure portfolio.
Its effect will be to allow corporate systems architects to deliver new systems based on flexible business models and changing business requirements – at this point Microsoft will become a close ally of its customers’ business planners.
Application fusion at Oracle
Oracle sits in the middle of the SOA platform contenders, as it has both an applications arm and a middleware division. And the latter is by no means merely an adjunct to the former – it brought in $1 billion of revenue in the last financial year, and according to the company’s own estimates, boasts slightly more customers than BEA has in total.
According to UK middleware director David Keene, that success lends Oracle’s SOA proposition a legitimacy lacking in SAP, the company to which it is most commonly compared due to their joint dominance of the applications market. Oracle’s commitment to open standards, he says, allows for just the kind of interoperability businesses are hoping to achieve from the SOA implementations.
“Oracle does not want to lock you into anything; we just want to give you the most value, and therefore adopt much of our stack,” he says. “We protect you by building it in open standards, and making it openly compatible with other software.”
Forrester Research rewarded what it calls Oracle’s “strong support for connectivity protocols” by naming its enterprise service bus (ESB) one of the market leaders in May 2006.
And there is more standardisation to go surrounding the ESB, says Thomas Kurian, senior vice president for Oracle Fusion Middleware. “The one piece that is not standardised is the service fabric,” he says. “What we are working on with IBM is a standardised service component architecture, which describes the way that services are tied into the bus.”
One overarching goal of Oracle’s middleware mission is to make SOA simpler. It is working with business process management suite vendor IDS Scheer to close the gap between the business analyst, who defines the business process, and the developer who programs and implements it in software form.
Similarly, Oracle is working on a declarative, visual programming platform to make coding for SOA an easier, and therefore for the business, cheaper thing to do.
All of this contributes to the impression of Oracle as the business-focused middleware provider. For Nicholas Gall, VP and research fellow at Gartner, Oracle’s SOA platform is still defined by its association with the company’s $3 billion business applications arm.
While it may be committed to interoperability of services, he says, that does not mean that its end goal is not to dominate the SOA stack from one end to the other.
“Any messaging you hear from them saying they are happy with being the platform for other vendors’ software is just window dressing,” he says.
BEA's 360 degree view of SOA
ANYONE attending BEAWorld in San Francisco in September 2006 can have been left in no doubt about the extent of BEA Systems’ SOA ambitions. What remains to be seen is how effectively it will deliver on its promises.
The promises are big ones. By the end of 2008, BEA 360° – the company’s new umbrella marketing term for SOA products – will provide, according to Rob Levy, BEA’s CTO, a unified SOA platform that is “more complete than anybody else’s in the market”, based on a BEA microService Architecture (mSA) middleware layer.
This is fighting talk. Whereas, BEA claims, most of its rivals are building SOA platforms that hide proprietary point-to-point middleware platforms in web services clothing, in mSA they will have the real thing: a truly loosely-coupled and granular “service network” or fabric that supports the dynamic discovery and deployment of IT assets in an event-driven environment.
The veracity of BEA’s claims won’t be known until it has completed the integration of new and existing products with mSA. This is scheduled to be complete in 2008, which doesn’t leave much time for BEA to perform a major recoding and integration effort around three very different core product sets: Tuxedo, WebLogic and AquaLogic.
Individually, each of these has good credentials to be a first choice pick in any custom-built SOA platform composed of best-of-breed elements. WebLogic, for instance, was recently judged the most scalable application server in an Evans Data Group survey, and the AquaLogic Service Bus is an equally respected ESB capable of straddling the worlds of Java and .NET. The AquaLogic Service Registry is another strong best-of-breed contender, now complemented by the AquaLogic Enterprise Repository (the metadata repository acquired from Flashline), and BEA’s BPM credentials were similarly boosted by the acquisition of Fuego in March 2005.
While the integrity and unity of the SOA 360° may or may not be proved in time, BEA still offers some major strengths and remains relatively well integrated overall according to analysts.
However, BEA has traditionally been costlier than its competitors. This has led its customers to pick and choose BEA products according to functional necessity, augmenting them with other commercial offerings or open source functionality. If BEA can realise its 360° vision on more than a superficial marketing level, it might be better able to justify its prices and overcome this historical purchasing pattern.