Evaluating Practical Tolerance Properties of Stabilizing Programs through Simulation: The Case of Propagation of Information with Feedback
Abstract
We simulate a stabilizing propagation of information with feedback (PIF) program to evaluate its response to perturbations. Under several classic execution models, we vary the extent of the fault as well as the system scale. We study the program's speed of stabilization and overhead incurred by the fault. Our simulation provides insight into practical program behavior that is sometimes lacking in theoretical correctness proofs. This indicates that such simulation is a useful research tool in studies of fault tolerance.