Back to results
US 11,455,122 B2Granted
Storage system and data compression method for storage system
AssigneeHitachi, Ltd.
InventorsNagamasa Mizushima, Tomohiro Yoshihara, Kentaro Shimada
Life of this patent
Estimating the term…
Abstract
Provided is a storage system in which a compression rate of randomly written data can be increased and access performance can be improved. A storage controller 22A includes a cache area 203A configured to store data to be read out of or written into a drive 29. The controller 22A groups a plurality of pieces of data stored in the cache area 203A and input into the drive 29 based on a similarity degree among the pieces of data, selects a group, compresses data of the selected group in group units, and stores the compressed data in the drive 29.
Claims14
A storage system comprising:
- a drive having a physical storage area;
- and
- a controller coupled to the drive and configured to receive data of a plurality of write requests from a host that are to be written into the drive, wherein:
- the controller includes:
- a cache area configured to store the data to be written into the drive, and
- the controller:
- groups a plurality of non-identical pieces of data stored in the cache area based on a similarity degree among the plurality of pieces of data into a plurality of groups,
- selects a group, which has an access frequency, on a group unit basis, which is smaller than a first set value and which has the similarity degree among data included in the group that is larger than a second set value and has a highest similarity among the plurality of groups,
- compresses data of the selected group in the group unit basis to form a compressed group,
- repeats the steps of selecting and compressing until a total amount of the compressed data for a plurality of the compressed groups satisfies a parity cycle,
- after the total amount of the compressed data for the plurality of the compressed groups reaches the parity cycle, generates a parity from the compressed data for the plurality of the compressed groups, and
- stores the compressed data for the plurality of the compressed groups and the parity in the drive,
- wherein for another group whose access frequency, on the group unit basis, is higher than the first set value, all data of the another group is maintained in the cache area without being compressed.
10 more elements
A data compression method for a storage system, wherein
- the storage system includes:
- a drive having a physical storage area;
- and
- a controller coupled to the drive and configured to receive data of a plurality of write requests from a host that are to be written into the drive, and
- the controller includes
- a cache area configured to store the data to be written into the drive, and
- the method, performed by the controller, comprises the steps of:
- grouping a plurality of pieces of non-identical data stored in the cache area based on a similarity degree among the plurality of pieces of data into a plurality of groups,
- selecting a group, which has an access frequency, on a group unit basis, which is smaller than a first set value and which has the similarity degree among data included in the group that is larger than a second set value and has a highest similarity among the plurality of groups,
- compressing data of the selected group in the group unit basis to form a compressed group,
- repeating the steps of selecting and compressing until a total amount of the compressed data for a plurality of the compressed groups satisfies a parity cycle,
- after the total amount of the compressed data for the plurality of the compressed groups reaches the parity cycle, generating a parity from the compressed data for the plurality of the compressed groups,
- storing the compressed data for the plurality of the compressed groups and the parity in the drive,
- wherein for another group whose access frequency, on the group unit basis, is higher than the first set value, all data of the another group is maintained in the cache area without being compressed.
11 more elements
Dependencies are inferred from the claim text.