International Business Machines Corporation

Daniel I Tan

Itzhack Goldberg

Gregory Tad Kishi

Joseph W Dain

Shazad Naviwala

A controller-implemented method includes reading, by a controller, a scrub unit containing a read unit and, in response to determining that the scrub unit is not marked as being scrubbed in a current scrub cycle, storing, by the controller, the scrub unit to a memory. The method also includes marking, by the controller, another scrub unit associated with the stored scrub unit as a scrub-pending scrub unit and reading, by the controller, the scrub-pending scrub unit in response to determining that a subsequent read matches predefined criteria. The method includes performing, by the controller, a scrub using the stored scrub unit and the scrub-pending scrub unit. A system includes a processor and logic integrated with the processor, executable by the processor, or integrated with and executable by the processor. The logic is configured to perform the foregoing method.

Performance enhanced data scrubbing

A computer-implemented method, according to one embodiment, includes: determining whether an error has occurred for each block of storage space in memory by sequentially evaluating each of the blocks of storage space. In response to determining that an error has occurred for a given block, the given block is scanned for an identifier at an end of data stored in the given block, and a trailer appended to the identifier is located. Metadata included in the trailer is further used to identify each additional block which corresponds to a same file as the given block, and a data scrubbing procedure is performed on the given block and each of the additional identified blocks. A result of the scrubbing procedure is also output in response to detecting an error in the given block and/or the additional identified blocks.

Leveraging distributed metadata to achieve file specific data scrubbing

A computer-implemented method, according to one embodiment, includes: receiving an access request for data in a first block of storage space in memory, and returning the data in the first block of storage space in response to the access request. An identifier at an end of the data in the first block of storage space is also located, and pointers included in a trailer appended to the identifier are used to identify additional blocks of storage space which include data having temporal locality with the data in the first block of storage space. The data in each of the identified additional blocks of storage space is further prepared for use.

Leveraging temporal locality to link files together and bypass accessing a central inode list

A system and method for data storage management is disclosed. The method includes determining, by an access tracking component, a quantity of access requests for at least one data block, and determining a quantity of current copies of the data block. The method also includes creating, by a duplication component, at least one additional copy of the data block when the quantity of the access requests exceeds an access request threshold. Additionally, a deduplication component removes at least one current copy of the data block when the quantity of the access requests falls below the access request threshold. The access request threshold can be a threshold number of access requests for the data block, the presence of an input/output bottleneck, or a given length of a read latency when accessing the data block. Further, data coloring techniques can be used to distribute current copies of the data block.

Growing community of inventors

Austin, TX, United States of America

Daniel I Tan