Documentação de Requisitos de Sistemas Legados: Uma proposta baseada na Engenharia de Requisitos Orientada a Objetivos

Abstract

Legacy systems are present in many organizations and are often responsible for managing their main and most valuable data. In addition to being obsolete, the documentation for these systems usually does not exist or is incomplete and outdated, the code is not structured, there is no support, there is a lack of specialized professionals, the business rules are under the domain/knowledge of few people, due to the lack of updates if make them vulnerable, among other aspects. These problems make the maintenance of these systems costly and risky, providing low-quality work that leads to the creation of even greater problems when a correction or improvement is made to the system, making it a nightmare for the teams that support it. Aiming to reduce these problems, organizations look for ways to modernize or even replace the legacy system, creating initiatives and projects that tend to be expensive and which historically have a high failure rate due to the complexity and problems mentioned above. In this context, as well as for the creation of a new system, it is understood that the first step in maintaining, evolving or even replacing the legacy system is to understand its objectives, impacts and functionalities through the system requirements. This work proposes an approach for initial documentation of the legacy system based on its objectives, seeking to identify objectives that are no longer met or that are partially met. It assumes that there is no prior existing documentation and that using the source code to extract requirements and generate artifacts is not viable. To document these requirements, it is proposed to use GORE (Goal Oriented Requirements Engineering) techniques. These techniques focus on representing stakeholders’ objectives, usually in a goal model, facilitating the requirements elicitation and documentation process. Among them, techniques such as i* (read iStar), NFR Framework, KAOS, GBRAM, among others, can be used to document functional and non-functional requirements of these systems. To support the identification of these requirements and objectives, studies point to traditional requirements engineering techniques such as: interviews, observation, protocol analysis as viable techniques for obtaining such information. From the models generated, it is proposed to use visual identification on goals that are no longer met, or that are only partially met, thus making it possible to make a more detailed assessment of the need for these requirements in a future update. A systematic review was carried out to support the proposal and update the community on which techniques, processes, methods and artifacts are currently used to extract requirements from legacy systems. The results of this RSL are used to prepare this proposal for documentation of legacy systems requirements