James Taylor's Decision Management

4. It will never work

There is sometimes an almost irrational fear that this stuff does not work.

• Old school programmers who resist any technology that does not require JCL
• Organizations where it's "just not how we do things here"
• "We are using methodology X and business rules does not fit"
• "We do agile programming and rules won't fit"
• Concerns that it will never perform well enough from folks who think assembler is the way to go and who never liked Java or C# either
• Concern that business rules can't be so great if they have not been heard of before
• A sense of hopelessness when it comes to getting the statisticians to adopt new technology
• And so on

Business rules management systems are a new technology for many organizations and do require different approaches. A business rules approach has been successfully integrated with leading methodologies like RUP and other UML-based approaches, as well as with agile methods. Business rules do require some additional steps and deliverables, especially to keep them separate from other requirements and mapped to use cases, but in general this is both straightforward and clear.

Most modern business rules management systems do not require complex new deployment environments, working well on Java and/or .NET and/or COBOL depending on the product. The leading ones are well behaved in terms of their deployment characteristics, data access, security and so on. Increasingly modern analytic tools are also focusing on how to deploy the resulting models as well as on how to build them and this is making deploying these models easier and less impactful on the IT architecture.

Performance of business rules management systems has improved enormously recently. With options to use compiled-out sequential execution where that makes sense and new and improved versions of the Rete algorithm (like Rete III) to improve inferencing performance, most customers find that their rules systems are, like their other systems, data-access or user-response bound. The rules and models run just fine.

Business rules have been, to some extent, an "under the radar" phenomenon. Now however major vendors like Microsoft and Oracle have created business rules-shaped "holes" in their development platforms, analyst firms like Gartner, Forrester and IDC are covering the market and standards bodies like OMG and W3C are developing rules standards. Meanwhile the growth in the use of predictive analytics for competitive advantage has been well documented in magazines like the Harvard Business Review.

This stuff works, the vendors can show plenty of good examples and the analyst firms and standards bodies think there is enough there to cover it. Don't get left behind now…