What it is ?
MINT is a software solution designed to correlate metrics from different software development life cycle in order to provide valuable recommendations to different stakeholders impacting the software development process. It considers the different measurements collected by the MEASURE platform as events occurring at runtime. The correlations are designed as extended finite state machines (EFSMs) allowing to perform Complex Event Processing in order to determine the possible actions (recommendations) that can be taken to improve the diverse stages of the software life cycle and thus the global software quality and cost.
Benefits
The basic idea behind MINT approach is to specify a set of correlation rules based on the knowledge of an expert of the software development process. These rules can rely on one or different sets of metrics (seen as inputs) and allow to provide different recommendations (seen as outputs) to different kinds of actors:
- Actors from the DevOps team: Analysts, designers, modellers, architects, developers, tester, operators, security experts, etc.
- Actors from the management plan: product manager, project manager, responsible of human resources, responsible of financial issues etc.
The tool is automatic and triggers recommendations according to the default correlation rules specified at the beginning. Now the set of correlation rules is small (10 rules) and needs to be extended. The tool is fully integrated to the MEASURE platform and can manage different measurements collected by the different measuring tools. More work is to be done to write new correlation rules or generate them from a previous data set.