Web services stack attack

It is day two of the IBM European Software Symposium 2003 in Munich, one of the largest conferences of its type in the European calendar, and the largest briefing room available has been set aside for a strategy session on web services.

But surprisingly, when Jason Weisser takes the floor, only a handful of people are in the room. Others have been lured away by rival sessions on mobile and collaboration technologies and the chance to get an early lunch. It is a small sign that web services, the most hyped technology of 2001 and 2002, has lost much of its lustre.

The session doesn’t begin well, either: “I have no idea what web services are and I’m proud of that fact,” says Weisser. This is a little surprising. He is, after all, the director of worldwide services for web services. But Weisser explains: “There is much, much more to web services than the standards [that dominate much of the discussion and early implementations today]. Web services is a few years away from making any sense. We don’t know what is to come. It will be interesting to see what web services looks like when it hits puberty.”

Weisser’s opinions may not sound that dramatic, but he is articulating a view that is fast coming to dominate the thinking of the leading strategists at some of the world’s most powerful technology companies – although it has largely fallen out of fashion among disinterested observers.

That view is that web services is a revolutionary, disruptive technology that will have far-reaching and unpredictable business consequences – just as many people were saying back in 2000 and 2001. And that means that those who now argue that web services is only a protocol, or a set of standards for linking applications, are taking a view that is much too narrow.

To be fair, the excitement does not centre on web services itself – which might be defined as a set of common technical standards for more easily linking applications and services. It centres on the wider concept of how all software applications, components and services, big and small, past, present and future, can be made to work together more easily, more cheaply and more effectively.

That concept was first described by the Gartner Group – as the Service Oriented Architecture (SOA) – in 1995. But it needed a combination of standards, including web services, before it could really work.

“WS is an iteration of a computing or a modelling paradigm that will become dominant over the next 20 years,” says Weisser. “That paradigm is the software oriented architecture.”

The SOA is not a fashion or a fad, nor is it a technology. It may be viewed as the coming together of the philosophies of web services, the J2EE environment and Microsoft’s .NET, and as the enabling architecture for the much anticipated ‘revolution’ in business process management (BPM).

Some analysts say the impact of SOA on the technology industry and its customers will be as great as that of the Internet. It will be long term, structural and irreversible.

But there are at least two critical differences. First, for the corporate customer, SOA should not require the huge up front investments that, ultimately, undermined and even discredited the first wave of the Internet business revolution. In fact, as long as the right choices are made, the SOA should lead to the simplification of IT strategies and, ultimately, to reduced or at least stabilised IT spending.

And second, whereas the Internet revolution sparked the creation of tens of thousands of new technology companies, the impact of SOA on the IT business will be quite the opposite. For SOA, read simplification and rationalisation.

It’s not the economy

These are challenging times in the software business. Many suppliers have suffered well-publicised financial problems, mostly related to the well-understood economic problems of the software industry, its rapid expansion and subsequent contraction.

But that is only half the story. Outside a handful of very, very large suppliers, there is scarcely a software executive in Silicon Valley or beyond who, in recent months, has not spent a portion of their days, and probably their nights, weighing up strategic options that are either tough, or risky or both.

It would be wrong to ascribe the atmosphere of uncertainty, even crisis, in the software industry directly to the emerging new paradigm. It is, for example, simply too early for the spending patterns of customers to have made an impact, in spite of the widespread use of web services in tactical projects.

“Only a minority of organisations are using web services in an SOA today. We only found three or four people who are deeply into SOA.” says David Sprott, chief executive of the consultancy CBDI and author of a detailed customer study on the subject.

But there is an atmosphere of strategic uncertainty that cannot be separated from the factors that created the opportunity for the SOA. Even many profitable and stable suppliers are wary of how the changes will play out.

For most customers, the medium and long-term gains will be substantial and measurable. But not for all: those caught with the wrong systems, or who misread the trends, will share in some of the pain of the suppliers.

Five big drivers are already having a substantial impact: Functional consolidation; Supplier reduction; Standardisation; Commoditisation; and open source development.

Although these factors overlap and interrelate, each is significant enough in its own right to cause disruption among both suppliers and corporate customers.

Functional consolidation

Early in 2003, Alfred Chuang, the CEO of BEA, bounced through the dry ice and onto the stage at the BEA eWorld conference in Florida, to declare “a major inflexion point of the history of enterprise computing”.

Why? Because from that point on, he said, with the latest tools from BEA, developers could integrate their existing applications (for example, commercial packages such as SAP or Siebel, or older, legacy systems) with the new ones they are developing. It meant, he said, “the convergence of yesterday and today”, and that from now on “all integration is development and all development is integration”.

All of this made for good marketing, and competitors were quick to rubbish both the hype and the substance. Many pointed out that BEA had tried much of this before, with limited success.

But very few people have questioned the general point: at least three sets of technologies are becoming so closely aligned and intertwined that the boundaries between them are now blurred. And those three technologies – application servers, development tools and suites, and integration brokers, represent multi-billion dollar industries.

That convergence of function can be seen clearly in the product releases and strategies of the suppliers: Microsoft, for example, has components for development (.NET Visual Studio), for application services (.NET and Windows) and for integration (Biztalk). And not only are these very tightly integrated, but parts of Biztalk are being re-engineered into Windows itself.

BEA is doing likewise with its WebLogic 8.1 product suite, which now includes development, deployment and integration services. Oracle now provides functions in all three areas.

And, of course, there is IBM, supplier of the Websphere integration and deployment suite. Its purchases of integration supplier CrossWorlds in 2001 and development and modelling tools supplier Rational in 2003 led one impressed analyst to remark: “This is beginning to look suspiciously like a strategy.”

Several close alliances within the industry further underline the point. Application server supplier Jboss, for example, has allied with integration specialist WebMethods and development tools supplier Borland.

This consolidation of function has clearly rattled many suppliers in the independent software industry – and their customers and investors. The obvious strategy of IBM, Microsoft, Oracle and BEA is to build a comprehensive stack that controls and integrates all the parts of the SOA.

And it is not clear where it will stop. All these companies already provide, or are likely to provide, portal software, business analytics, business process management and even complete business applications. And from the other end, SAP is increasingly looking to join the platform suppliers – some analysts expect the company to break its cautious stance and make a major acquisition.

Supplier reduction and spending

At the same conference where he announced his historical inflexion point, BEA CEO Alfred Chuang allowed himself a public grumble with the press and analysts who cover his company. “I thought if we got to $1 billion [annual sales], we wouldn’t be asked these questions anymore,” he complained, pointing to his company’s impressive financial performance. “We have produced one of the most spectacular P&Ls [profit and loss accounts] and balance sheets in the industry.”

Those questions were: Are its products in danger of being commoditised? Will competition from IBM, recently merged with Rational Software, hurt sales? Is it losing share? Will Oracle buy it? How much damage will open source software do to its business?

Some of this may have been token scepticism. But only days before the conference, respected analyst group AMR Research had warned: “BEA is surrounded by heavyweights. Everyone wants a piece of the business.” And Forrester Research had released a report a few months earlier saying BEA wouldn’t be a long-term survivor.

The point is clear enough. If BEA is under threat, then how much more worried must the smaller vendors be? Ask those same questions of executives at almost any smaller software company supplying tools for developing, integrating or managing software, and their replies will frequently be tinged with anxiety.

The service oriented architecture can accommodate, but does not require, multiple control points or multiple platforms. And research across the world has shown that where customers have an opportunity to simplify and reduce the number of suppliers, they will. “The cost savings of buying one co-ordinated infrastructure solution can be dramatic,” said a recent report from the Giga Group.

“Because of cost pressures, many organisations are going to have just one or two suppliers of component technologies,” says Marie Wieck, vice president responsible for worldwide business process integration for IBM. “But it is too early to say how it will play out.”

That is potentially devastating for smaller, niche suppliers – hence the very obvious and urgent strategic repositioning by, for example, the integration software vendors. But it also means that customers will choose just one or two of the big three to five platform vendors. In other words, even the likes of BEA or Oracle could be marginalised at some accounts.

Again, research clearly shows that one of the chosen environments will usually be Microsoft’s .NET – and the other will be J2EE. And J2EE suppliers such as IBM and Oracle will win many of the big accounts because of their strengths in areas such as databases and operating systems.

This explains the pressure on BEA, in spite of its performance. It needs still more volume and market presence. “Its preferred model is world domination. The second option is co-existence,” says Phil Dawson, international programme director for infrastructure at the Meta Group. But for smaller companies, the preferred option is co-existence, the second a satisfactory exit.

Standardisation The goal of standardisation is to take complexity and cost out of IT. But one company’s costs can be another company’s revenues.

Nowhere is this more evident than in the field of integration – where there are five pure play suppliers (Tibco, Mercator, SeeBeyond, WebMethods, and Vitria) with combined revenues of more than $1 billion. Web services standards provide a base layer of interoperability that enables software package suppliers, and end users, to replicate some of the functions that the specialist integration suppliers provide.

And it is not only the web services standards that address this: J2EE application servers now have the JCA (Java Connector Architecture) and Microsoft will build a similar capability into future versions of Windows server.

Some analysts think the effects can already be seen. In the first quarter of 2002, the combined revenues of the five pure play integration vendors fell 12%. And although this is largely because of economic factors and strategic repositioning, some customers are now wary of investing in specialist integration products.

Mike Parks, the CIO of Virgin Mobile, is one. He recently implemented a complicated multi-application SOA-style architecture with no middleware. “Three or four years ago, you would have had a middleware layer. But we are of the mind: Why have an integration tool separate from the application server?”


Competition drives down prices. Globalisation drives down prices. And so does standardisation. And in the case of the new software architecture, all three of these factors are at play – and will be for many years to come.

This is good news for customers, but it means there is a long-term, structural problem for many suppliers of enterprise software. But once again, it hurts the smaller, specialist suppliers the most, and the big infrastructure suppliers the least.

Take, for example, Microsoft. As Peter Bell, head of the company’s development and platform evangelism group in the UK points out, BizTalk, the integration product, is the most expensive product in the Microsoft catalogue – but it is also the cheapest in its marketplace. That is because of Microsoft’s high volume, low price business model.

IBM has a different approach: it will discount its core platform software, such as the DB2 database or the Websphere application server, in order to win long-term account control. This means more services, a long-term flow of maintenance revenues, and the opportunity to cross-sell some of the hundreds of other products in its portfolio.

Oracle, too, has been heavily discounting its 9i application server to win market share and market acceptance of its entire applications platform stack.

The Giga Group recently described the competitive situation in application servers as “a roaring buyer’s market”. This is a description that, with the changes that the SOA will bring, might be applied to large swathes of the software industry for several years.

Open source development

The issue of open source software has divided the software industry – not least because it is one area where Microsoft has been hurt. In a recent memo, Steve Ballmer, the CEO of Microsoft, warned his staff that: “Non-commercial software products in general, and Linux in particular, present a competitive challenge for us and for our entire industry.”

But there is much more to open source than the Linux operating system. In 2002, for example, there were 2 million downloads of JBoss – the open source application server (prompting one BEA executive to say, “Never was a piece of software downloaded by so many, and used by so few.”) In February 2003, when it was revealed that General Electric, a BEA customer, had set up a JBoss competency centre, BEA’s shares plummeted.

Similar open source initiatives are underway in databases, XML products, collaboration tools and content management systems – to name but a few.

The challenge that open source software presents to the software industry is not, directly, linked to the emergence of new, open software architectures – except in one crucial respect: the ability to locate and access processes and functions, and then plug them into what once would have been proprietary environments, will dramatically extend the reach of open source software.

By adhering to standards, it will be possible to plug open source components as services into, for example, a .NET or Websphere environment; and analysts expect there will be many more fine-grained open source application functions available.

The next decade

Where does all this leave the software industry? The twists and turns of the ebusiness and Internet revolution, and the failure of so many previous software revolutions, demonstrates how unpredictable the technology industries are.

Yet analysts agree on at least these points: many software functions are converging on single, standardised platforms; these will all use SOA architectures and web services, and either .NET, J2EE or both; all other architectures will be wiped out; smaller vendors will struggle to survive – and so will some fairly large ones; and, provided customers make the right choices, IT will be much cheaper and more flexible than hitherto.

Avatar photo

Ben Rossi

Ben was Vitesse Media's editorial director, leading content creation and editorial strategy across all Vitesse products, including its market-leading B2B and consumer magazines, websites, research and...

Related Topics

Web services