INTERVIEW: EAI is dead; long live Web services

Cape Clear Software is one of a handful of startup companies creating development and deployment environments around Web services. As the CEO of Cape Clear, Annraí O'Toole is betting that a startup is better suited to leverage a new style of computing than an established player. In an interview with InfoWorld Editor in Chief Michael Vizard and Test Center Director Steve Gillmor, O'Toole explains why traditional EAI is dead and how Web services will inevitably triumph.

Q: How did Cape Clear come about?O'Toole: The basic inspiration for setting up the company was that we saw that there was going to be some merger between the traditional middleware environments, the Internet, and XML. When we set the company up in 2000, we didn't know what that was going to be. When Web services came along, we didn't have to explain it all from first principles. Essentially what we're doing is applying Web services to the problem of integration. We believe that Web services can essentially address all of the problems in integration, or 80 percent of the problems that the vast majority of the market wants. The key thing we like about Web services is that they democratize a lot of the middleware stuff. It's transforming the economics and the technology of integration.

Q: How will people differentiate between where to apply Web services and where to use something else?O'Toole: I think it breaks down into an 80/20 rule: Twenty percent of the integration problems will remain complex, requiring expensive proprietary solutions. But the vast majority of integration is going to be achieved by Web services. To the extent which that replaces what the traditional EAI [enterprise application integration] guys are doing, well then we're going to replace them.

Q: So from your perspective, a lot of existing EAI vendors will fade away?O'Toole: They have to rebuild their products from scratch to be Web services. [We describe] the solutions that the incumbent vendors are coming up with for Web services as putting lipstick on the products. They are taking out this dusty old product and giving it a nice coat of lipstick. If they look at Web services, they can substantially reduce the cost and the skills needed to do integration. They're not going to buy the lipstick only to find out that they need to bring with that 150 consultants to install and run this proprietary integration software. There is a separate Web services design center that isn't trying to approach the problem from a traditional EAI solution or the traditional J2EE [Java 2 Enterprise Edition] environment.

Q: What impact are companies such as Microsoft having on the EAI space?O'Toole: What Web services are doing is commoditizing the lower levels of the stack. Microsoft has effectively started that process. The higher levels of the stack are now helping people orchestrate the various systems that they want to tie together, defining those business processes and extending that over the Internet. It's pretty certain that over the next few years you're going to see Web services do extremely well at internal integration. But ultimately, this thing will be about exposing these business processes out to the outside world. And that will happen. When and exactly what kind of business models is going to drive this out to the outside world, no one knows that yet.

Q: What's limiting Web services today?O'Toole: There's quite a lot of nasty reality that has to be dealt with today. Although these standards are an enormous step forward, there's a lot of work required in getting them all to work together. We've spent a huge chunk of our engineering time in two directions. One is interoperating with the whole .Net environment. And the .Net environment is bigger than SOAP [Simple Object Access Protocol] and WSDL [Web Services Description Language], it's this whole .Net framework, the various servers, BizTalk, and the development environment. Similarly, we connect on the back end to Java. Making sure that all this stuff seamlessly interoperates is a non-trivial piece of work, but this stuff doesn't come out of the box all plugging together.

Q: How does your approach differ from other vendors?O'Toole: We don't see that customers are going to buy, say, four different products. They're just not going to go do that. They want to buy it all in one. We've kind of coined this phrase of an integrated Web services environment where you get to do all tasks from the one environment. The whole thing about Web services is democratizing this hard-to-use rocket science middleware and bringing it to the masses. They are typically people who are closer to the business problems but who know a bit about computing. They're not going to be afraid about writing a bit of script or using some kind of graphical programming environment. Those are the sets of people that we're ultimately after. Heretofore, there's just no way on earth that you can turn any kind of moderately competent programmer into a CORBA developer or a J2EE developer or an MQSeries developer. Those technologies have just been beyond their skills range. There traditionally has been always this huge divide where the average human beings typically haven't been well served by software vendors. Microsoft has been really the most successful vendor in that space with Visual Basic. The most important thing for us is to have a complete offering in the box so you can take it out and it all fires up. You've got everything there that you need to get going. You don't need to download other stuff.

Q: What's your take on the Microsoft vs. Java debate as it relates to Web services?O'Toole: Microsoft really sat down and spent some time thinking about what happens when you combine a ubiquitous networking environment like the Internet with software delivery and software packaging. That's what brought them to Web services. In my view, they deeply understood and analyzed that and built a set of products on the technology strategy around that, whereas the Java community hasn't. So for all of the moaning of the various guys about how they're doing Web services, the response so far in the Java community has been to add yet more APIs into the J2EE container as if it needed any more APIs. I mean the thing is already bizarrely complex as it is. So adding in yet more APIs and making this already complex thing even more burdensome and complicated is the wrong answer. It's driving it further and further from the reach of those mere mortals. The response so far from the non-Microsoft world about building an architecture for Web services hasn't been that impressive.

Q: How important is asynchronous support for Web services going to be?O'Toole: Asynchronous Web services will be one of the most predominant usages of the technology. With enterprise architects looking at this stuff, that's the thing that they get most excited about. We've got a lot more to do in terms of building that in. We offer a kind of basic, limited asynchronous functionality in of the box today. During this year, we will have strong integration with other environments. As I describe them, they're the drains that people have already installed inside their organizations. There's no way on earth that you can go around to them and say "Rip up all your drains." You'll get shown the door pretty quickly. So integrating with all those environments is kind of essential. Web services do not offer any compelling alternative to publish and subscribe multicast technology. It just doesn't come near it. There's a whole range of customers for whom that's an absolute requirement.

Q: How important are workflow languages going to be?O'Toole: The standards committees are working to try and reduce them down to at least one standard. A key part of this whole Web services standard is the ability to orchestrate them together, to describe all the sets of interactions between all these entities that are out there. You need a common declarative way to do that. Either they'll get all standardized and there will be agreement, or one of them will become more relevant than the other. That will be a good thing for the industry as well.

Q: What impact has the economy had on interest in Web services?O'Toole: The way you might look at it is that for $5 million, rather than getting me two sets of things talking to one another, I can get 20 things talking to one another. I think that budgets are probably going to remain flat. Rather than buying a million dollars worth of some proprietary EAI tool, you can spend $100,000 and get the same amount of technology that's cheaper to deploy and faster to learn. Those things become pretty compelling.

Q: So what do you expect to see happen by the end of this year?O'Toole: We're seeing enterprise architects looking at all this Web services stuff. During this year you're going to see a number of kind of major accounts who are going to stand up and say "We've adopted a total kind of enterprisewide Web services architecture." That, for me, is the very necessary thing to occur in 2002.

Q: So what's the big take away on Web services?O'Toole: Web services are part of an inextricable movement in the software industry, which is to turn software into a service and make it more like a utility. That's been running throughout the industry for as long as it's existed. Web services are just the latest iteration in that movement. The customers are clearly thinking about it in those terms. They know in the back of their minds that's where this stuff is ultimately leading them.

Join the newsletter!


Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

More about Microsoft

Show Comments