When Bill Gates, the founder and chairman of Microsoft, addressed over 150 press and analysts at a seminar to explain his company’s strategy for web services in July, he adopted a disarming tone of contrition.

The roll out of .NET, the core part of Microsoft’s two year old ‘bet the company’ web services strategy, he conceded, had not gone quite to plan. In fact, he said, there had been a series of missteps, the core ideas and technologies had been badly communicated, and there was an enormous amount of work yet to do.

He went on to give his company a series of grades for its .Net execution. For development of basic building-block software – C; for the roll out of Microsoft’s new ‘software-as-a-service’ concept – “still a lot to be done”; in changing users’ experience of the software – Microsoft had done an “incomplete” job; security – also “incomplete”; in .Net Services, “we made a couple of missteps”; in branding – “too many products carried the .Net name”.

In the weeks that followed, Microsoft executive managers joined the chorus. “Web services is something that you can present in four slides, but we presented it in forty,” said Phil Cross, Microsoft UK’s .Net developer marketing manager, admitting customers have been confused.

The company even confesses to fumbling its ISV (independent software vendor) strategy: “We had a huge focus on ISVs two years ago. It was successful, but in true Microsoft style we said ‘job done’ and moved on. We took a bit of focus off it,” says Cross.

All of this has, according to Gates, inspired executives to take a hard look at what they are trying to do with .NET. “[We are] doing a bit of a reset,” Gates said.

A reset? The word was immediately seized upon by Microsoft watchers. “When Bill Gates says ‘reset’, it is with the explicit meaning of pressing the reset button when the PC has crashed,” wrote one.

In the software business, resets are not uncommon. But .Net is not just any project: “This is about creating software that connects information, people, systems and devices. It is transforming everything we are doing. It is the holy grail,” Gates told analysts. This echoed a speech he had given when .Net was launched. Then, he said, Microsoft was betting the company on the technology.

If the whole .Net project is being redrawn, or is in some kind of serious strategic difficulty, then Microsoft is in the process of losing its position of leadership in the software industry.

But Microsoft is not saying that .Net is in any kind of trouble. In fact, it is actually saying it will redouble its efforts to make sure that .Net is the most successful set of web services – by far. And most industry analysts think that Microsoft’s softly-softly tone is merely a tactic. For all its problems, most say that Microsoft is ahead, even far ahead, in web services. And while execution is a massive engineering task that Gates says will take at least half a decade, one of its biggest worries is actually appearing too successful. The anti-trust authorities are always watching.

.Not?

Scott McNealy, the quotable CEO of Sun Microsystems, once joked that .Net is ‘vapourware’ and should be called ‘.Not’.

But that was a while back. In March 2002, Sun, which sells servers, development tools and management software for web services, asked a federal court to stop Microsoft from tying its Windows operating systems into its .Net technology. “If they are allowed to continue their bundling with .Net, they are going to control the on-ramps and the exit ramps to the Internet,” McNealy said at the time. His views are widely shared and more legal challenges are expected.

McNealy and Gates don’t agree on much, but they do both share the view that web services – whether it is based on .Net or on the ‘J2EE’ open standards architecture – is the future for software and for the Internet. Customers will be able to easily access software, services, data and business processes from across a network; the age of standalone packages is coming to an end, and will be replaced by subscription services.

Even in its infancy, web services technology, and .Net in particular, has made a deep impression. Early adopters of the technology are gushing. Take this comment from JP Ragaswami, global CIO at Dresdner Kleinwort Wasserstein: “Web services is by far and away the only way to go… highly visible, expensive, glorious, celebrated failures are far less likely to happen.”

Or this endorsement of .Net from Jon Higgins, Tesco.com’s head of ecommerce development: “It makes strong financial sense. We can develop an application once and then reuse it… web services is enabling us to get an immediate return on investment.”

Microsoft knows how much customers love it. But it also sees its enthusiastic adoption of web services standards as a way of strengthening its portfolio, finding new customers for already-existing products and services, and for getting existing customers to buy more.

For a supplier that is already the runaway leader, that is clearly threatening to rivals – especially those who are established suppliers of enterprise solutions. “The next phase [of .Net] is clearer [to Microsoft]. Become a strategic supplier to the enterprise by making every desktop a gateway to corporate data and collaboration,” says Ted Schadler, an analyst with Forrester Research.

But the extent of that vision is making many in the industry nervous. Microsoft’s vision, despite its fumbled execution so far, is so extensive that there are few companies that are will not find themselves competing against the giant in some way. Many fear they will forced into Microsoft-compatible niches.

Embrace, extend, extinguish

At the core of their concerns: standards. Almost every supplier – including Microsoft – has emphasised that, for web services to work, open standards must be used. That includes, for example, the alphabet soup of SOAP, XML, HTTP and UDDI (see Web services crib sheet).

But Microsoft’s record on standards is patchy. For example, it chose its own COM (common object model) technology over the industry standard Corba object request broker technology; in languages, it developed its own dialect of the Java language, J++, and later its own equivalent, C#; in XML, it has fallen out with other members of the ebXML standards group.

Now, rivals point out that its tight embrace of open standards comes with a big caveat: applications built using .Net development tools will be able to interact with applications based on non-Windows operating systems, but they must nevertheless themselves run on a Windows server. And not every business wants to base key corporate applications on Windows servers.

Online analyst service the451 points out, “J2EE [the Java-based web services architecture] is fundamentally open, and .Net is essentially closed. Microsoft will be the sole provider of the core.Net technologies.”

The main reason for this is that Microsoft has developed a technology called the common language runtime, or CLR. Similar to a Java virtual machine, it is essential to running .Net and only runs on Windows. (Although a small company, Ximian, is developing a version to run on Linux).

It is primarily this component, along with its C# language, that sets Microsoft apart from rivals such as Sun, IBM, HP and BEA, all of which are committed to using the Java language, the Java virtual machine, and the J2EE component architecture.

Apart from that, all the big software and web services platform suppliers – Microsoft, IBM, Oracle, Sun, BEA – offer or intend to offer a family of development, management and integration products that do similar jobs and which, in many cases, are interchangeable. The main differences are ones of emphasis, features and routes to market.

Microsoft, in defending itself against charges of being outside the mainstream, argues strongly that it supports every web service standard, and that includes the Java language. Its goal, it says, is not to dominate the architecture, but to sell application services, development tools and, of course, operating systems. No-one, it points out, who develops web services using .Net will be cut off from interoperating with any partner who uses J2EE.

Industry leaders believe that, in spite of the squabbles, the early success of web services is very largely due to the fact that workable standards, such as SOAP and XML, are supported by everyone. Rick Bergquist, chief technology officer of PeopleSoft, uses the analogy of Ethernet in the early 1990s. “Ethernet standards allowed the explosive growth of networking; it made the whole industry grow. Vendors across the board can see the network effect of getting people talking to each other [using web services].”

Developers the key

The key to future of .Net, as distinct from web services, lies in the take up of the Microsoft tools that differentiate the company from the rest of the suppliers. These include, for example, the C# language and the .Net Studio product. And here, a familiar pattern is emerging.

Microsoft’s development tools (such as Studio .Net) have a distinct edge over Java-based tools in smaller organisations and in niches of large organisations, where cost and ease of use are especially important.

“If you compare the ease of use of Microsoft tools and Java tools, Microsoft wins hands down, and that’s a big advantage for them,” says Uttam Narsu, VP for XML and web services at researcher Giga Information Group.

But Java stalwarts stress that Microsoft development tools can’t produce the robust, high-end applications that many larger enterprises need. “There will be key applications which need high transaction rates and links to lots of different systems. Microsoft can’t deal with yet,” says Ian Doyle, senior architect at BEA Systems.

While Microsoft dominates the low end, then, surveys of large corporations have usually shown that a substantial margin prefers Java over .Net as a corporate development environment.

This is the same fault line that runs through the entire IT business – whether in the user community or among suppliers. Microsoft Windows and its related tools dominate the low end, and Unix and J2EE dominates the high end. In the same way, but to a lesser extent, Intel machines dominate at the low end, and specialist processors dominate at the high end.

This fault line, however, is beginning to move – and that is makes the stakes much higher. There are two dynamics at work.

First, it is a matter of function, price and skills. Almost all large businesses, for example, already use Microsoft’s servers and development tools as well its ubiquitous desktop products because they do the job, they have skills in-house, and because the price is right. On this front, the balance of power between Microsoft’s development tools and their Java-based rivals has been changing gradually in favour of Microsoft – and it could change faster as Microsoft server and development-tool technology continues to mature.

But it could equally move the other way: open standards, better and cheaper J2EE tools, and, crucially, open source components, could weaken Microsoft’s grip on the desktop and on the reseller community.

But there is a second, crucial dynamic: web services is just part of the wider move towards server-based computing, or software-as-a-service. This model means the client environment is much less important than the control of the server-based infrastructure. And that means Microsoft needs to develop better server-based products – products that directly challenge those offered by its big J2EE platform rivals. (The trend also explains why Microsoft introduced its controversial Software Assurance subscription-based licensing scheme).

Who is winning so far? Surveys show that .Net and the Java J2EE community are closely matched. A Merrill Lynch poll in March 2002 found .Net almost head-to-head with Java when CIOs were asked which they would choose as the foundation of their web services strategies. And while a November 2001 survey by Evans Data found that 55% of developers plan to use Java, and 35% .Net, that numerical advantage reflects Microsoft’s dominance at the cheaper low end. It may not translate into control of the all important server.

Gartner Group, meanwhile, expects C# to become an important language, with a million professional developers using C# worldwide by late 2005, compared with just over 2.5 million each for (Microsoft) Visual Basic and Java by that time.

All of these figures, however, show just how powerful Microsoft is: the surveys show Microsoft taking maybe a third, maybe half, of the market. That leaves a few dozen suppliers to fight for the rest.

That assumes, however, that the ‘fault line’ will change dramatically. Factors that could destablise it are Microsoft’s bold move into enterprise applications, which is already alienating powerful business partners such as Siebel; and IBM’s increasingly strong story in web services integration technology.

And, of course, there is always the possibility that the new paradigm will spring its own surprises. “There’s a new Oracle or Dell out there that will emerge from web services. Microsoft will do well, but it won’t be the big winner,” says Barry Morris, CEO of web services and integration software provider Iona.

Analysts think that the battle for support among ISVs, whose support can make or break products and technologies, won’t be won conclusively.

Microsoft may be building an ‘ecosystem’ around .Net, but it’s one dominated by one big fish, and the others must beware. “There’s fear in the ISV community about getting too close to Microsoft. That’s one reason why Java is so popular among ISVs. In the Java world, you get a feeling that there’s an ecosystem, and that there’s more variety,” says Giga’s Narsu.