All aboard the WS-* standards express

IBM and Microsoft in the driving seat

Warning: train

Myths and legends It seems there is a disquieting trend in IT: concepts are getting steadily vaguer, and claims harder to verify.

Take web services, for instance. The very name is disingenuous. They are services of a kind, but they don't have much to do with the web. Their key protocol is SOAP, which stands for Simple Object Access Protocol. Well, it is a protocol, all right. But it isn't simple, and it doesn't access objects.

Such is the vagueness, that if you ask five people to define web services, it's been said you'll get six different answers.

The idea of XML as a networking payload actually goes back to 1996, when it was working through the W3C. Almost immediately companies like webMethods and Bowstreet pioneered techniques for XML-based middleware. And Hewlett-Packard devised a promising architecture called E-Speak but characteristically forgot to tell anyone about it.

Then, in early 1998, Dave Winer of UserLand wrote his famous blog entry predicting that the next big thing would be "RPC using XML over HTTP". Winer got Microsoft interested, and its increasingly complex protocol became known as SOAP while his simpler version was called XML-RPC.

It was Microsoft's June 2000 announcement of .NET marked a turning point in the company's history. Rather than trying to "embrace, extend, and extinguish" the internet and the web, Microsoft switched to "embrace, extend, and excel".

Instead of confrontation, why not set the trends, and perhaps even own the development environment? After all, it's axiomatic that platform sales follow applications, applications follow developers, and developers can readily be wooed - if you spend enough money. Accordingly, an early white paper pledged that: "Windows will offer the best environment to create and deliver web services, while Windows-based clients will be optimized to distribute web services to every kind of device".

But why did Microsoft decide to collaborate with IBM when it came to handing down web services standards? Not long before IBM had formed a series of alliances - for instance Kaleida, OpenDoc, and Taligent - aimed at competing with Microsoft, and the two giants were still serious rivals in 2000.

Could their sudden rapprochement have been impelled by mutual hatred of Sun Microsystems? Or was IBM recalling Sun Tzu's advice to: "Keep your friends close, and your enemies closer"? This scary partnership began with a joint submission of SOAP 1.1 to the W3C in May 2000. IBM and Microsoft then announced UDDI (September 2000, jointly with Ariba) and WSDL (March 2001, with many other companies).

Soon there were so many SOAP-related specifications in the WS-* stack that the vendors had to found the Web Services Interoperability Organization (WS-I) just to let users know which configurations of the many standards would actually work in practice - and, just as important, how to make their web services interoperable.

As early as 2001 IBM's Bob Sutor made it clear that Big Blue expected special treatment from standards bodies.

What it boiled down to was that he didn't want any unnecessary delays while amateurs and academics bickered. In fact, he seemed to be telling consortia like the W3C and Organization for the Advancement of Structured Information Standards (OASIS) that they should eat what they were given by industry leaders like IBM and Microsoft. The attitude was: "Just for you, here is a part-baked standard that we made earlier. Just form a committee, talk it over for a couple of months, and then rubber stamp it. Trust us, we know best".

That was presumably why, after submitting SOAP and WSDL to the W3C, they chose to give UDDI to OASIS instead. Later, they alternated between the two, allocating some specifications to OASIS and others to the W3C.

With web services we have entered a new era in standards setting. Instead of giving up control to a vendor-neutral standards body, IBM and Microsoft kept the initiative themselves. That way they stayed in charge, but they retained plausible deniability; and they can keep their secrets until it suits them to go public. Best of all, if anything goes wrong the standards consortia are there to take the blame.®

Tom Welsh is a senior consultant with Cutter Consortium's Enterprise Architecture advisory service. Tom has been following OMG and its specifications since 1992.

Sponsored: 10 ways wire data helps conquer IT complexity