A cloud-based operator training system includes a snapshot management architecture, which provides a hybrid system for generation of control system level scenarios and system-state snapshots, and which can improve the fidelity of a training simulation. By implementing the simulation system on a cloud platform, the system can generate a large and growing set of snapshot files representing various control states and corresponding process states. These files can then be leverage during operator training sessions to yield high fidelity simulated system operation.