| Abstract |
Knowing the provenance of a data item helps in ascertaining its trustworthiness. A key challenge when developing provenance aware systems is deciding what provenance should be captured during the execution. Leveraging full-system execution trace logging and replay, we present a 4-stage methodology and a system which allow the complete decoupling of provenance analysis from the original execution. This enables selective and flexible provenance analysis using progressively heavier instrumentation, instead of relying on a single pre-determined type of analysis. |