Interview: GM's CTO outlines Web services strategy
- 12 May, 2003 22:00
Tony Scott, chief technology officer of the information systems and services organization at General Motors Corp., spoke with Computerworld about the automaker's plans for Web services technology, the expected benefits from them and the possible risks GM hopes to avoid. Excerpts from that interview follow: Do you feel Web services have been overhyped?
I would not write off any technology that the likes of Microsoft and Sun and other large companies like IBM have invested hundreds of millions if not billions of dollars or so in R&D. The likelihood is pretty good that it's going to hang around and be something significant.
At GM, we're interested in it from a number of different perspectives. We've all lived through the embryonic stages of the Web and all the EAI [enterprise application integration] tools and the big application packages, so we don't see Web services as the panacea, the end-all, be-all that's going to replace everything else. But it does fit, or appear to fit, in the Web development space in particular, where we can take certain kinds of activities that we're embedding in each and every application we build today and externalize them as a service and only do, for example, maintenance and repair on that service in one place vs. in hundreds of applications.
Can you cite an example?
Obviously, every car has a vehicle identification number. And there are dozens if not hundreds of applications in GM that use vehicle identification numbers for something as a part of our normal business process. There's a whole set of business rules around vehicle identification numbers -- good ones, bad ones, when the vehicle was made, its repair history, all kinds of things. And today, we have to embed lots of business logic and rules in every single application that uses vehicle identification numbers.
One of the things we are working towards is creating a vehicle identification number Web service that in effect will encapsulate all of the logic and business rules and so on -- let's call them the big rules around vehicle identification number -- so that you don't have to support and maintain that in hundreds of applications.
The positive benefits of that on quality, on consistency and so on, I think you can probably imagine. It's pretty significant. And there are, in a large company like GM, dozens of examples -- whether it's people or parts or locations or airline tickets or whatever -- where there's redundant logic, redundant business rules, redundant overhead and maintenance and support costs associated with everyday things that we deal with in our business.
Have you settled on a development environment for Web services?
We're unlikely to pick just one, given the breadth of GM. In a company the size of GM, it's really hard to pick one and then make a case that that's going to meet all the needs of everybody at GM. But we do like to pick a few.
Do you feel encouraged by the potential for Web services to help you connect applications written using Java and Microsoft's .Net technologies?
Yes. In our proofs-of-concept work, we've taken simple ideas and we've done exactly that. We've married up what's shipping today among the various vendors, not totally without problem, but I'm more encouraged than I've been in a long time in terms of basic interoperability.
GM made a big commitment to EAI technology from SeeBeyond Technology Corp. Do you foresee Web services having any potential to replace those kinds of investments?
I don't see them as a core replacement. The dividing line that I see is that we've still got lots of legacy applications. I think any organization more than a few years old has these. Some of these are pretty heavy transaction-oriented systems that do the heavy lifting of our day-to-day business. And you need a high-power integration engine to do those.
So we're unlikely to rip that out and try to replace it with Web services. Now, it could be you still use the EAI tool deep in the bowels of the company, but you have some Web services front ends, if you will, or technology as the interface to some new applications that you're developing. I think that's very plausible, and I think all the EAI vendors recognize that, in terms of their support for Web services standards and so on.
Do you have any Web services in production?
Yes, we have a few internal. Where we tended to put them in production generically are in places like our GMAC unit, our financial services arena, where it's not running 60 factories around the world. ... Generally what they're doing is application-to-application interface in a smaller-scale environment.
What benefits will GM gain from Web services?
I think the benefits will come not just from Web services but more from a broader context within which Web services fit, which is all-around reuse -- better initial architecture understanding. If you think about using a Web service, it forces an architectural discipline into your environment that you may not have had. Of course, you can implement anything sloppy. But done the right way, it should act as a catalyst for a more architected approach to implementing IT.
What are your greatest concerns as you head down the Web services path?
One is fragmentation. There's a sordid history in the technology world of everybody trying to get a little leverage over somebody else by developing proprietary extensions or vendor-specific add-ons to the core technology. In general, those have been bad, because they don't end up being sustainable over time and that costs companies like GM a lot of money.
The other area that worries me is that, while I think there's a lot of benefits to standardization and interoperability, from a security and risk standpoint, hackers or people with evil intentions have an opportunity to exploit a flaw and create the equivalent of an electronic weapon of mass destruction, if you will. I think we in the IT industry have been somewhat protected by the level of diversity that we've had in our systems. And the more standardized we get, the more interoperable we get, we may expose some of these vulnerabilities that can be more fatal than we've experienced in the past.
Have you seen the sort of progress that would make you comfortable enough to do external Web services?
I think we're several months to a year away before I'd feel comfortable in that space. There are two issues for me. One is absolute security, and the other is the scalability of the security model. For GM, scalability is as much an issue as the absolute security. There are lots of things that can be done very securely. They just don't happen to scale very big. And so this trade-off between scalability and absolute assurance is a tough one. It's not an easy problem.
GM outsources its IT operations. Will GM outsource Web services development?
Yes. All of it will be. We're 100 percent outsourced. There's no team of people here that are coding, building and implementing business applications. Our staff here primarily do the business requirements, the architecture and design work and then the actual building and support and operation is 100 percent outsourced.
Do you think the outsourcing philosophy is conducive to Web services?
Yes, because we think one of the advantages of this model is it really enables you to move fast. We can go out and select an outsource vendor who has the most up-to-date skill sets and we don't have a large internal workforce that we have to worry about. We can cherry-pick what we need and apply it rather quickly.
Will you dictate the development environment to the outsourcer?
It depends on the area. I think, in some areas, we'll be prescriptive in terms of what we want them to use mainly for support and ongoing maintenance reasons. In other ones, it might not make that much difference and we'll let the market decide what's the best technology to use at any given point. -- Computerworld (US online)