Towards hybrid model persistence

Change-based persistence has the potential to support faster and more accurate model comparison, merging, as well as a range of analytics activities. However, reconstructing the state of a model by replaying its editing history every time the model needs to be queried or modified can get increasingly expensive as the model grows in size. In this work, we integrate change-based and state-based persistence mechanisms in a hybrid model persistence approach that delivers the best of both worlds. In this paper, we present the design of our hybrid model persistence approach and report on its impact on time and memory footprint for model loading, saving, and storage space usage