Delta Compressed and Deduplicated Storage Using Stream-Informed Locality.

For backup storage, increasing compression allows users to protect more data without increasing their costs or storage footprint. Though removing duplicate regions (deduplication) and traditional compression have become widespread, further compression is attainable. We demonstrate how to efficiently add delta compression to deduplicated storage to compress similar (nonduplicate) regions. A challenge when adding delta compression is the large number of data regions to be indexed. We observed that stream-informed locality is effective for delta compression, so an index for delta compression is unnecessary, and we built the first storage system prototype to combine delta compression and deduplication with this technology. Beyond demonstrating extra compression benefits between 1.4-3.5X, we also investigate throughput and data integrity challenges that arise.