发明名称 Optimizing and enhancing performance for parity based storage
摘要 A mechanism is provided for optimizing and enhancing performance for parity based storage, particularly redundant array of independent disk (RAID) storage. The mechanism optimizes a repetitive pattern write command for performance for storage configurations that require parity calculations. The mechanism eliminates the need for laborious parity calculations that are resource intensive and add to IO latency. For repetitive write commands that span across the full stripe of a RAID5 or similar volume, the mechanism calculates parity by looking at the pattern and the number of columns in the volume. The mechanism may avoid the XOR operation altogether for repetitive pattern write commands. The mechanism may enhance secure delete operations that use repetitive pattern write commands by eliminating data reliability operations like parity generation and writing altogether.
申请公布号 US9405625(B2) 申请公布日期 2016.08.02
申请号 US201514670844 申请日期 2015.03.27
申请人 International Business Machines Corporation 发明人 Fiske Rahul M.;Gunda Kalyan C.;Jones Carl E.;Patil Sandeep R.;Roy Subhojit
分类号 G06F11/10 主分类号 G06F11/10
代理机构 代理人 Tkacs Stephen R.;Walder, Jr. Stephen J.;Bluestone Randall J.
主权项 1. A method, in a storage controller, for optimizing and enhancing performance for parity based storage, the method comprising: receiving from a host device a repetitive write command and an associated pattern to be written a plurality of times to a portion of a parity based storage volume, wherein the parity based storage volume employs block-level striping with parity and wherein the repetitive write command has an associated flag; determining, by the storage contoller, whether the flag is set; determining, by the storage controller, whether a size of data to be written by the repetitive write command spans a full stripe size of the parity based storage volume; determining, by the storage controller, whether a stripe unit width of the parity based storage volume is a whole number multiple of a size of the associated pattern; responsive to the storage controller determining that the flag is not set and the size of data to be written by the repetitive write command spans the full stripe size of the parity based storage volume and the stripe unit width of the parity based storage volume is a whole number multiple of the size of the associated pattern, optimizing parity calculation for the repetitive write command by the storage controller, wherein optimizing parity calculation comprises setting parity for a given stripe within the portion of the parity based store volume to be the pattern or zeroes based on a number of columns in the parity based storage volume; and writing the pattern with the parity to the portion of the parity based storage volume based on the optimization.
地址 Armonk NY US