US Patent 10078451 Deduplicating data based on boundary identification

Among other things, we describe a technique used in data deduplication that includes receiving a single data file designated to be written to a file storage system configured to store data in the form of blocks. The technique also includes identifying boundaries between portions of data within the single data file, and providing an indication to the file storage system to allocate blocks to the single data file based on the identified boundaries. Each block is allocated to, at most, one of the portions of data. The technique could also be used with objects instead of files.