Columns

The Microsoft/IBM axis

• By David Chappell
• September 1, 2002

I have spent much of my career involved in one way or another with standards for communication. In particular, I've focused on protocols above the transport layer, an area where broadly supported standards for application-to-application communication have been hard to come by. Without these standards, networks are nowhere near as useful as they might be. Once they exist, though, the world opens up like never before.

SOAP-based Web services are opening the world today. But SOAP isn't enough. Basic communication is essential, but real systems require much more. And through the WS series of specifications, Microsoft and IBM are working together to create the standards we need to make this a reality. These specifications are the future of Web services.

The joint Microsoft/IBM specs in this series include WS-Security, an essential addition to the basic SOAP protocol; and WS-Inspection, a simple way to discover the WSDL definition or other information about a Web service available at a known location. When you read this, more specs will likely be available. Each firm has talked publicly about its interest in transactions, and in reliable messaging and other critical aspects of building an effective, interoperable Web services environment.

And let's face reality here: Microsoft and IBM own Web services. While the two firms eventually cede control of the technologies they create to more formal standards bodies (SOAP goes to the W3C, while WS-Security goes to OASIS), the truth is that the creation of those technologies is in the hands of just two vendors. Is this a perfect situation? Maybe not. Is it the one most likely to produce widely accepted Web services standards? Probably. Is it going to change anytime soon? Clearly not, so we may as well get used to it.

Microsoft refers to the WS series of specs as the Global XML Web Services Architecture (GXA), and initially published GXA specs on its own. The truth, though, is that these specs didn't really matter until IBM's name also appeared on them. Microsoft's original WS-Security spec, for example, has now been superseded by a jointly produced version with the same name. Whatever your opinion might be of the specs' technical merits, the simple fact that two software giants agree on them makes them all-but-guaranteed to succeed.

Wire-level interoperability, with competition reserved for the implementations, is the model that has made our modern distributed world work. After years of at least partially resisting this model, Microsoft and IBM seem to have bought into it. The two vendors will surely continue to compete fiercely, as the ongoing battle between .NET and IBM's Java-based WebSphere shows. This is a good thing: Competition forces the combatants to either improve or die.

And from the evidence so far, both companies appear committed to interoperability. Having gone so far down this path, each side now has a great deal to lose if they deviate from it. The world we're headed for appears to be one with two dominant programming environments -- .NET and Java -- but one common set of communication protocols. As more WS specs appear, and implementations of those specs work their way into products, multiplatform environments will become easier to deploy. And since neither .NET nor Java is going away anytime soon -- most large enterprises will have both -- this interoperability is critical.

Of course, once these two giants have defined the standards, every other vendor (this means you, Sun) is effectively forced to fall into line. The likelihood that even Java application servers will use Web services to interoperate -- the Java specs themselves don't define much in this area -- is a great irony. Who would have imagined that technologies whose creation owes so much to Microsoft would ultimately solve the problem of interoperability between Java-based products?

Which vendor -- IBM or Microsoft -- is the major force behind the WS technologies? Are they equal contributors? I don't know and I don't care. All that matters is that both have committed to implementing them in an interoperable way. I've been waiting my entire career for application-to-application standards. Now that they're finally arriving, I couldn't be happier.