Abstract: A medium, system, and method are disclosed for implementing an in-memory inode cache. The cache stores inodes that include respective file metadata. The system includes a replacement module that prioritizes the inodes for replacement according to retention priorities for the inodes. The system also includes a prioritization module that is configured to increase the retention priority of a given inode in the cache in response to detecting a cache hit on the given inode and that the given inode meets one or more longevity criteria.

Abstract: A medium, system, and method are disclosed for implementing an in-memory inode cache. The cache stores inodes that include respective file metadata. The system includes a replacement module that prioritizes the inodes for replacement according to retention priorities for the inodes. The system also includes a prioritization module that is configured to respond to a file close operation by determining whether the file is unlikely to be reopened. If so, the prioritization module responds by decreasing the respective retention priority of one of the inodes, where the one inode stores file metadata for the closed file.

Abstract: A computer-implemented method for redundant data storage may include identifying a storage system. The computer-implemented method may also include identifying a unit of data to be redundantly stored. The computer-implemented method may further include identifying an integer that represents the unit of data. The computer-implemented method may additionally include identifying a set of pairwise coprimes. The computer-implemented method may further include generating a set of shares, with each share including a reduction of the integer modulo a corresponding pairwise coprime in the set of pairwise coprimes. The computer-implemented method may also include redundantly storing the unit of data by storing the set of shares on the storage system. Various other methods and systems are also disclosed.