On-disk file format for a serverless distributed file system

A method comprising:

• segmenting a sparse file into multiple blocks, the sparse file containing at least one non-data block that contains no substantive data;
• differentiating the non-data blocks from data blocks of the sparse file that contain substantive data;
• computing hashes of each of the data blocks to produce block hash values;
• encrypting the data blocks using their corresponding block hash values as encryption keys to produce encrypted data blocks;
• creating an indexing structure to index individual blocks, the indexing structure containing first leaf nodes for each corresponding encrypted data block and second leaf nodes for each corresponding non-data block, the first leaf nodes containing an access value formed by encrypting the block hash value for the corresponding encrypted block using an access key and a verification value formed by hashing the corresponding encrypted block;
• and
• setting the second leaf nodes to a first binary value.

One or more computer readable storage media comprising computer-executable instructions that, when executed, direct a computing device to:

• segment a sparse file into multiple blocks, the sparse file containing at least one non-data block that contains no substantive data;
• differentiate the non-data blocks from data blocks of the sparse file that contain substantive data;
• compute hashes of each of the data blocks to produce block hash values;
• encrypt the data blocks using their corresponding block hash values as encryption keys to produce encrypted data blocks;
• creating an indexing structure to index the non-data blocks and the encrypted data blocks;
• and
• deallocate portions of the indexing structure that reference the non-data blocks.

A component in a device in a distributed file system in which files are stored across multiple distributed computers, the component comprising:

• a segmenting module to divide a sparse file into multiple blocks, the sparse file containing at least one non-data block that contains no substantive data;
• a control module to differentiate the non-data blocks from data blocks of the sparse file that contain substantive data;
• a hash module to hash each of the data blocks to produce block hash values;
• a cryptographic engine to encrypt the data blocks using their corresponding block hash values as encryption keys to produce encrypted blocks;
• and
• an index builder to create an indexing structure to index individual blocks, the indexing structure containing first leaf nodes for each corresponding encrypted block and second leaf nodes for each corresponding non-data block, the first leaf nodes containing an access value formed by encrypting the block hash value for the corresponding encrypted block using an access key and a verification value formed by hashing the corresponding encrypted block, the second leaf nodes being set to a first binary value.

A method implemented at least in part by a machine comprising:

• segmenting a sparse file into multiple blocks;
• differentiating non-data blocks in the sparse file that contain no substantive content from data blocks in the sparse file that contain substantive data;
• creating an indexing structure to index individual blocks;
• deallocating storage of both the non-data blocks and portions of the indexing structure that reference the non-data blocks that contain no substantive data, wherein the sparse file is reduced in size;
• computing a hash of each of the data blocks to produce block hash values;
• and
• encrypting the data blocks using their corresponding block hash values as encryption keys to produce encrypted data blocks.