Matthew Hotle is a Group Vice President and Research Group Director at Gartner. His experience includes technical and business management, project leadership and management, software engineering, quality assurance, testing and training.
Matthew began his talk by referring to himself as a certified cynic,which is a good way to look at Web Services especially in regard to legacy transformation. In the 80s, he wrote some of the legacy code still being used today, and informed the audience, There is not enough SOAP in the world to clean up some of the stuff I wrote!
Matthew said that to find the potential in Web Services it is necessary to have a strong sense of reality regarding what you can do, how you can do it, and the work needed to make it happen. It is not a simple matter of just wrapping it. Reusing business processes can be done in only two ways: re-engineering or wrapping. Wrapping works best when the programs are loosely coupled and cohesive (the degree to which a program does one thing and one thing only). This is important to remember when choosing programs that are candidates for Web Services. Loosely coupled programs means you dont have to send the interface and business logic with a program when it has to be used elsewhere. COBOL programs with more than 5000 lines of code are candidates for re-engineering, not wrapping.
Matthews conclusions about Web Services are that legacy extension solutions can be used to wrap current functions as Web Services, sometimes very quickly and easily. Isolating key business functions and transforming them to newer architectures yields the greatest value, and restructuring will yield finer-grain business functions for Web services. He pointed out that you always have to separate the presentation logic from the business logic.
Matthew said that a year ago, Gartner asked 500 people when they were planning to use Web Services, and 65% responded that they already were. Since this was way ahead of the adoption rate, the conclusion was that most were defining Web Services as anything they were running over the net. Matthew defines Web Services as:
- Loosely coupled, dynamically bound (true dynamic binding is 7-10 years away)
- Platform and programming language neutral
- Using ubiquitous Internet protocols (WSDL, SOAP, UDDI)
- Described, published, located and invoked over a network
Matthew then broadened his own definition by saying that Web Services are, loosely coupled, coarsely grained, business processes delivered in software over the Internet.He added that the heavy liftingfor open standards for Web Services has increased their use rapidly.
Matthew said we are moving into a new phase of business processes, called the real-time enterprise, where we can move from concept to concrete in a very short period of time. The average time for a piece of software to go from concept to functional use is about a year. But we dont have that much time anymore. We need to do it in three or four months. Matthew said that we are going from a transactional mode to using and wielding our technology as a strategic weapon. The stakes are high, for everyone. Matthew related a recent conversation on reuse when someone asked him who would make the reuse work. Matthew told him that he must make it work, because if he didnt someone else would take his job and make it work. The business demands it.
The attributes of this new way of business are:
- Externally focused business model
- Significant new constituents
- Dynamic supply chain interaction
- Fundamentally new business model
- New external sources of AD
Matthew gave an example. A friend of his in the Chicago area works for a company selling the Saddam trading cards. The business owner took his idea to market in less than two weeks. The company now sells 8000 cards an hour on the Internet. Using the new model, they were able to take the idea to market with a new product in less than two weeks. They were able to contact new clients they didnt even know about before. This new business model allows you to create, get in, and get out of a market in a period of months.
But there are problems:
- Performance problems (no service level agreements)
- Wrong granularity of services (tightly coupled software)
- Security concerns (lots of worries here)
- Incompatible runtime environment
- Cultural hurdles
The potential, however, is there. The expected ROI for this is 3.5:1, over a period of five years, according to Matthew, and this goes up to 6.5:1. This is a lot of payback, but it requires some substantial changes.
Matthew said that Gartner sees clients taking major advantage of Web Services for application integration. They are building bottom-up interfaces to existing applications using Web Services. He said that this process is more an assembly process than building process. Matthew pointed out that only 20% of development budgets are spent on new development. The rest is spent on legacy systems, with legacy being defined as anything that is currently in production. If you want to improve your business processes and your applications development processes, you should be spending your time on the remaining 80%.Older assets can be rejuvenated using Web Services.
Matthew then showed a slide with the five areas of legacy-to-Web Services products and listed the major vendors in each. These are:
- Business rules identification
- UI/Data extension adapters
- Legacy transformation
- Web Services
- Legacy Web services
The presentation ended with a series of Gartner magic quadrants, showing the Gartner rankings of vendors for legacy understanding, presentation integration server, and legacy transformation.
His overall recommendations are:
- Make business strategy decisions first. Is reuse an appropriate solution?
- Use legacy extension tools for applications with low performance concerns.
- Consider tactical XML tools to begin initial efforts.
- Transform applications to more modern environments in limited cases.
Matthew made the case for Web Services when he said, Web Services are going to be the great equalizer. They allow us, through a standards-based stack, to tie things together. They allow us, in the short term, to do application integration, some application interoperability, and some supply chain work. The services-oriented architecture is open enough so you can start today building service-oriented applications using your existing stuff and not do all that much. It is a fairly high pay-back and a fairly low-cost mechanism to make this work. And finally, you cannot cost-justify getting off the mainframe in four to six years.