Large volumes of data are being created, disseminated, and processed on the Grid. To maintain accountability while the data is transformed by multiple parties, a consumer must be able to check its lineage and deem it trustworthy. This project is building filesystem support to transparently generate and certify Grid data provenance, and a corresponding tool to validate it. The system's ability to provide assurances will scale despite the fact that provenance metadata grows exponentially, is generated in multiple, independent administrative domains, and must be verifiably ordered in the absence of certified, global, synchronized timestamps. The approach is trading certification space and verification time, uses out of order validation of certificate chains, and embeds forward-secure temporal witnesses in the metadata. Its deployment will let scientists efficiently query and verify the lineage of data processed in external Grid domains.
