If you ask most analysts today what the next big thing to come from the computer industry will be, they'll probably tell you, "Web services." They'll point to the fact that Microsoft, IBM, Sun and Oracle, along with all the major software players, are involved with Web services to some extent. It seems like every other week, someone is proposing a new initiative or forming a consortium around the technology. Web services are all the rage.
It's surprising, then, that they're also so widely misunderstood. Ask three people what a Web service actually is, and you're liable to get three different answers. Once all the hype and marketing is stripped away, it can be hard to tell exactly what's left. Which is odd — if Web services are the next big thing, shouldn't it be easier to tell just what the heck they are?
What They Ain't
So many misconceptions abound, in fact, that it may help to first clear the air about what Web services aren't. First of all, they're not a way to run your desktop applications over the Web. Forget about that idea of renting your word processor by the month and having the code delivered to you over the network in real time — that's not going to happen anytime soon.
For others, the mention of Web services calls to mind the ultimate privacy nightmare, where companies amass giant databases of personal information about you based on the services you use. However, while ideas like that have been floated in the past, they really have little to do with the underlying technology. In fact, many Web services won't need any personally identifying information at all.
Much of this confusion arose as a result of the early hype surrounding Microsoft's .Net products. But though Web services are important to the new platform, .Net is a companywide initiative that encompasses many other technologies as well. Microsoft's confusing marketing blitz has sometimes made it difficult to tell which buzzwords belong with which. You might even think Redmond was the only company offering Web services, though nothing could be further from the truth.
It doesn't help that Sun and the other competing vendors have been so closely tracking Microsoft's every move. Sun has been busy offering up alternatives even to Microsoft's most vilified technologies, like the Passport single sign-on authentication service. To an outsider, observing this market must be a little like watching Spock play that 3-D board game on the original Star Trek. You can tell it's chess, but you can't figure out whether either side is actually getting any closer to winning.
Under the Hood
In reality, the basic premise behind Web services is fairly simple. They're meant to be an easy way to build networked applications from modular, distributed components. You take a library of code, put it on the network somewhere and give other software the tools to access that code as if it were running on the local machine, and — voila! — a Web service (more or less). The "Web" part is even something of a misnomer. It stuck because these services are often accessed using HTTP, the Web-server protocol.
This idea has a long history, dating back through technologies like CORBA, Microsoft's DCOM and even Apple's OpenDoc. What's significant this time around is that Web services are completely standards based and platform agnostic. The protocols that let applications identify and make use of Web services — including SOAP, WSDL and UDDI — are all based on XML, itself an open, industrywide initiative. Applications that access a Web service don't even need to know what kind of computer the service is running on to make use of its functions.
But if network-delivered software hasn't taken off, what are these services really good for? Maybe more than you think.
While running a word processor from a Web server doesn't make much sense, a lot of other complex applications already do work that way, including e-commerce storefronts, intranets and reporting software. These types of programs often need to communicate with mainframes and other so-called legacy systems — systems that weren't designed with the Web in mind and that often use proprietary interfaces — to get the data they need.
Putting a platform-independent Web service between the Web server and the legacy system is one way to solve this communication problem. The Web service can act as a middleman between the two systems, translating messages from one into a language that the other can easily understand.
Since Web services use a standardized network interface, the same service could perform similar duties for systems outside the company as well. It could help connect a company's clients or vendors to an internal database, for example, for easy sharing of records. Most analysts believe this is likely to be the largest area of growth for this technology in the near future.
New Dog, Old Tricks
Performing this type of middleware function is a valuable use for Web services, and they'll probably gain a fair amount of traction in the industry for that reason. But it's a far cry from the radical sea change in the way we approach computing that some of the hype has painted them to be — which brings us back to our earlier discussion: what Web services aren't. For one thing, they aren't, perhaps, as revolutionary as the industry press has trumped them up to be.
Consider CDDB, an online database of song-title information for compact discs. CD-playing and MP3-ripping software can use a standard interface to query this database over the Internet and get results back, no matter what operating system that software is running on. The software doesn't need to know anything about how the CDDB data is stored, or what internal operations were necessary to retrieve it. CDDB was designed long before the modern Web-services protocols, but it works along similar lines — proof positive that these ideas weren't exactly born yesterday.
For all the rhetoric pumped by Sun and other Web-services proponents, on close examination this technology just doesn't seem that groundbreaking. Web services are one way to solve some problems the industry has, like interapplication communications — but there are other ideas out there that could conceivably do the job just as well.
Even some Web-services supporters question the level of sturm und drang that's been whipped up around them. Among the doubters is Oracle CEO Larry Ellison, who spoke on the subject at last year's Oracle OpenWorld conference. "The application-integration problem has nothing to do with Web services," he scoffed. Citing his company's strong track record of producing products compliant with the Web-services protocols, he nonetheless referred to the idea that Web services could be a panacea for the industry's interapplication-communications problems as "monumental idiocy."
Served Half Baked?
Of course, a big part of Larry Ellison's job these days is being a loudmouth. I'm sure that, to him, the best solution to the application-integration problem would be for everyone to standardize on Oracle applications. Still, he does have a point. Those of us who follow the tech industry — the press especially — love to get excited about new ideas. It's only natural. But while computing has given birth to some really radical, paradigm-shifting ideas in the past, Web services isn't really one of them.
The emperor's new clothes? I wouldn't go that far. Even Larry Ellison agrees that Web services have a value, or else his company wouldn't have built support for them into so many of its products. But though they seem to be the next big thing right now, it just doesn't seem likely they'll have the same kind of long-term impact as, say, Linux, or the Web itself.
But you know what? Maybe that's okay. We've heard a lot of crowing about the Web and software industries over the last few years. But with the implosion of the dot-com economy, a lot of the investment buzz has shifted to other fields, like biotechnology. Now that software companies are somewhat out of the limelight, maybe we'll begin to see a return to an era where what we call "innovation" relies on pragmatic solutions to problems, rather than hype.
For us, your average, everyday computer-using public, tomorrow's next big things probably won't be as star studded and spectacular as we'd like them to be. But maybe they will help us get some actual work done, and start the industry moving forward again. I wouldn't call that a letdown — in fact, I'd say it's a pretty good thing.
