Operating System Support for Replay of Concurrent Non-Deterministic Shared Memory Applications
Mark Russinovich · 2007
Repeatable execution is a crucial component of cyclic debugging, performance monitoring and fault tolerance based on journaling. For example, in cyclic debugging bugs are located by repeatedly running an application and then replaying executions that exhibit erroneous behavior in order to isolate the cause of the problem. In fault tolerance, programs running on a system that has failed can be recovered by starting them from a saved earlier state and bringing them up to the state that existed at the time of the failure by recreating inputs and other external events.