Mechanisms are provided to maintain deduplication data in native file formats. Files, including entities such as volumes and databases, are analyzed to identify components suitable for deduplication. These components suitable for deduplication are delineated into chunks and identifiers are generated for each of the chunks. The identifiers are used to reference the chunks in deduplication dictionaries that provide locations indicating where deduplicated chunks are stored. The components in the files are replaced with file handles or stubs that applications can use to access deduplicated data. Applications can continue to perform operations on the files as though no deduplication has occurred.