A generative framework for IoT migration of legacy systems

点击翻译按钮获取中文摘要

Internet of things (IoT) applications provide insights to companies about their products and services to improve their decision-making processes, which explains the growing interest in adopting IoT technologies. Yet, companies face several challenges when migrating their legacy applications to the IoT. Consider a hardware store that wants to migrate their inventory management system towards the IoT. Using sensors, the store wants to locate its products in real-time. To achieve the migration, we must evolve the legacy system from a call-and-return architecture (CRA) to an event-driven architecture (EDA). In the CRA, components communicate through synchronous calls and new data is provided through user inputs. In the EDA, events enable asynchronous communication between decoupled devices and components, and new data is provided via events of various sources (e.g., sensors that signal product attribute changes). This CRA to IoT EDA migration is not trivial when there is a high number of heterogeneous devices and, possibly, multiple Cloud platforms. Existing IoT EDA platforms can help in the migration. Yet, choosing the appropriate platform and performing the migration towards this platform is not straightforward because they provide different capabilities in terms of event streaming and management, performance, reliability and security. The goal of this research program is to build an approach composed of a methodology and tools to help with the migration of legacy CRA systems towards IoT EDA. The proposed methodology to achieve this goal relies on empirical engineering and model-driven engineering. In particular, we will carry out empirical studies to identify common and dissimilar features of IoT EDA platforms, and to build a catalog of recurring changes-called IoT migration patterns- that a legacy system must undergo in a CRA to IoT EDA migration. We will leverage model-driven technologies: (1) to abstract common IoT EDA platforms and automatically derive a platform description model for a given platform, (2) to codify the identified IoT migration patterns in an operationalizable form, (3) to map components of the legacy system to the appropriate IoT migration patterns and (4) use these mappings to semi-automatically support the refactorings embodied in the IoT migration. The outcome of this program will be of great value to Canadian enterprises seeking to benefit from the IoT technologies. It will contribute to a better understanding of IoT platforms and to improve the IoT migration process and reduce its cost. This research program has also potential for a significant impact on research. It will augment current architectural knowledge with IoT technology solutions, define IoT migration patterns and support them through semi-automatic refactorings. The HQP involved in this program will be trained in model-driven technologies, IoT technologies and IoT migration patterns. They will be in high demand by software companies.