Abstract :
[en] Engineering the requirements of the right transaction model (right with respect to the business goals of the application to be supported) is a hard task since it involves the critical choice of the right degree of ACIDity, that is the right selection of requirements in terms of Atomicity, Consistency, Isolation and Durability, which altogether are fundamental to ensure dependability and, more specifically, reliability. Up to now, this task is definitively not supported by a process. This thesis provides a novel process, called PRISMA. PRISMA is a Process for Requirements Identification, Specification and Machine-supported Analysis that targets transaction models. PRISMA is helpful as a prism in the identification of fundamental and constituting properties of transaction models to achieve, as a result of the PRISMA process, a correct and valid requirements specification.
The main idea behind PRISMA is that transaction models may be considered as a product line and that variabilities and commonalities may be identified to distinguish similarities and differences among "products". Specifically, PRISMA is conceived for engineering the specification of a transaction model by placing the effort in revealing its Atomicity, Consistency, Isolation and Durability requirements, which represent the variabilities of the product line.
PRISMA contributes on one hand in increasing quality, in particular dependability, and on the other hand in reducing time to market and cost by intensifying reusability.
