News

Web Services—Pointing the Way

• By Gary Barnett
• March 12, 2001

Microsoft's .NET announcement, and the subsequent slew of 'me too' announcements by the company's arch rivals hails what promises to be a huge sea change in the way we build and deploy business processes.

But remember that object databases promised to "transform the way we store and use data". We know that there is a huge difference between a 'promising technology' and a 'mainstream' technology, and the road from 'cool' to 'mainstream' is littered with the bones of technologies that paid the price of promising way too much way too soon.

The good news is that web services has some powerful friends—Microsoft, IBM, Sun, and HP to name but a few. It also seems as if the time is right. E-commerce, the need to adapt business processes quickly, and the very real need to find effective ways of sharing those business processes with partners and customers all cry out for some means to be able to easily exploit existing business processes, and make them available to other people.

Most interestingly however is the appeal that web services are going to present to developers—the majority of whom would gladly swap the need to know the intricacies of IIOP and COM RPC for a straightforward and clear mechanism that allows them to build applications that interact with other applications over a network. This is after all the primary purpose of all those 'heavy duty' distributed computing technologies.

The business of making a service available to others via the Internet is actually relatively simple. From the CGI scripts of the early nineties to the richer XML based mechanisms of today the task has been straightforward: First you define a simple interface and then some means to call that interface and retrieve any output that your process might produce. Indeed, the very act of entering a URL follows exactly these steps.

By using an ubiquitous and simple protocol like HTTP with an ubiquitous and simple mechanism for formatting Data (XML), you have the seeds of a model for distributed computing that is functional and, above all, simple to use. Technical simplicity often draws scorn, but I'm a huge fan.

Whilst there are a slew of 'minor details' that need to be addressed by the owners of technologies like SOAP and UDDI, the fact that so many major players on the supply side of the technology world find web services compelling, and the fact that it could lead to a model of distributed computing that doesn't require a Ph.D. in nuclear physics to exploit, means that of all the 'Promising Technologies' I've seen for a while, this one seems likely to keep some of its promises.