RESUMEN

Many software development professionals and managers aren't fully convinced that programs to raise their organizations' CMM level will provide a good return on investment. Unfortunately, because of the scattered nature of published research on this topic, the results haven't been conclusive. An investigation of CMM level transitions for seven common performance metrics (including ROI) analyzed a database of more than 400 projects from 19 information sources. Each database record was a normalized "after-before ratio," which signifies improvement for a given metric and is independent from the studied organization's measurement methods. This analysis found that CMM program investments improved performance consistently for all seven metrics, despite the wide range of organizations and software development practices. A statistical analysis supports these findings.

Using the arrival theorem together with elementary facts regarding integrated tail distributions and length-biased sampling we obtain closed form expressions for the inter-output time and the cycle time distribution in cyclic, single-class Jackson networks. Corresponding expressions for the normalization constant and the throughput are also obtained.