Empirical Evaluation of Lift-and-Shift for Decoupling Drivers in Industrial Legacy Software: Lessons from a CGI Case StudyBest Paper Award - Industry Track
Industrial Operational Technology (OT) systems are specialised hardware and software that directly monitor and control physical processes and machinery in industrial environments. Legacy OT systems are difficult to evolve due to tight coupling to hardware, vendor libraries and external integrations. Although internal business logic often remains stable, even minor interface or library changes require full system rebuilds and extensive validation, connected to high maintenance and risks. In this context, this paper presents an empirical case study conducted in a real industrial setting at CGI and focused on a 20 year old C++ monolith. We show an incremental modernisation approach that isolates external interfaces from the legacy OT monolith and re-implements them in containerised environments using Docker. Our results show that the approach improves modularity, reproducibility and maintainability while preserving the legacy logic. The paper demonstrates that mainstream IT engineering principles, such as containerisation and virtualisation, can be applied to long-lived OT systems and without full replacement. Our paper offers practical guidance for gradual, rather than big-bang, OT modernisation under real operational constraints, including lessons on isolating the dependencies and interactions with the underlying hardware.