10.7 Summary

Model problems are an effective component-based software engineering technique for evaluating design contingencies. In this chapter, we developed a model problem to evaluate the feasibility of maintaining transactional integrity from COBOL to EJBs, using Micro Focus COBOL, WebSphere, and Oracle 8i. We created and demonstrated a model problem that satisfied the ending evaluation criteria for the model problem ”increasing our confidence in the viability of the design option.

As a result of this model problem, the system architect identified this design solution as the principal design contingency. Becoming the principal design contingency does not guarantee that the solution will be adopted, but typically the principal contingency will receive the most resources to further verify the viability of the approach and reduce design risk. For example, we identified some additional risks during the implementation of the model solution, including how well the COBOL types defined for interacting with Java map into the COBOL types used in the RSS. At the same time, we did little to verify the performance, robustness, and scalability of this approach. These attributes must be considered further.

Although the principal contingency is not always adopted, it is an important step nonetheless. As a result of this decision, other design solutions are potentially starved of evaluation resources. If, for example, we had selected the OTS approach, we may never have applied the resources to identify the mixed-programming language approach as a viable design option. Time is also an important factor. As time passes , the design becomes more entrenched in the principal contingency. This design solution then becomes a design constraint in other model problems and as engineering expertise is acquired in the requisite technologies. Eventually, the cost of replacing the principal design solution with a contingency becomes prohibitive.