Facets

Traditionally, if two files are considered identical, they are understood to be byte-by-byte equivalent. However, many files are semantically identical without being byte-to-byte identical. We assert that by using a contextual or implicit notion of alikeness, new recovery mechanisms, caching algorithms, security protocols, and backup tools can be developed with a reduced storage overhead penalty and robust guarantees regarding secure deletion. To that end, we have developed and prototyped the Facets system, which leverages filtered provenance data to store and recover files, cache equivalent sets to improve performance, securely delete files, and minimize the overhead penalty incurred when performing backups. We view the Facets approach as broadly applicable and extensible within the realm of storage.