Automated selection of software components based on cost/reliability tradeoff

Functional criteria often drive the component selection in the assembly of a software system. Minimal distance strategies are frequently adopted to select the components that require minimal adaptation effort. This type of approach hides to developers the non-functional characteristics of components, although they may play a crucial role to meet the system specifications. In this paper we introduce the CODER framework, based on an optimization model, that supports "build-or-buy" decisions in selecting components. The selection criterion is based on cost minimization of the whole assembly subject to constraints on system reliability and delivery time. The CODER framework is composed by: an UML case tool, a model builder, and a model solver. The output of CODER indicates the components to buy and the ones to build, and the amount of testing to be performed on the latter in order to achieve the desired level of reliability.