|
Presentation
Empirical Analysis of COTS Activity |
||||||||||||||
|
Empirical data has revealed that COTS based application (CBA)
development lifecycles are unique and differ from traditional software
development processes. Each project will vary considerably in the
particular amount of effort expended on COTS assessment, COTS
tailoring, and COTS glue-code development. As such, there are wide
variations in cost/schedule/quality factors, risk items, and project decision process profiles. Previous work has described these variations and provided a composable COTS decision framework that models how such variations emerge within the COTS application development process.
We expand on this work by elaborating the sequence in which COTS assessment (A), tailoring (T), glue-code (G), and custom development (C) activities are performed. These sequences provide a “genetic code” for a CBA development project and are useful in characterizing the uniqueness of a CBA development lifecycle and enabling tactical decision support such as identifying high-risk development patterns, effort and schedule planning, options valuation. Other applications include analyzing the effects of risks on COTS activity and validating the composable COTS decision framework motioned above. We present CBA activity effort sequences from 9 USC e-service projects and their relationship to reported project risks. In addition, we present an initial set of “anticipated” and “un-anticipated” CBA effort sequence patterns. Anticipated sequences are patterns that have been observed within our case studies and are rationalized with respect to COTS risk-reduction on a project. One would expect to see such patterns on a well-managed, successful CBA development project. Unanticipated sequences are patterns that have not been observed, or when they have been observed, have correlated strongly with high risk and perhaps project failure. |
|||||||||||||