The success and wellbeing of large organizations rely on the smooth functionality and operability of their software. Such qualities are largely affected by evolution events and changes, like software upgrades. In this paper, we are dealing with handling evolution events in data management systems. We consider a data-centric ecosystem that captures relational tables, views and queries (the latter are seeing as software modules that are either internal to the database, e.g., stored procedures, or

more »

... external software applications that access the database). We also consider policies dictating the response of a software module to a possible event. We investigate the impact of such events to the database and present a graph-based mechanism to control event propagation. We show that our mechanism terminates and that every database construct is annotated with a single status, regardless of the sequence of messages that the node receives.