Fast data recovery from HDD failure

• US 8,639,969 B2
• Filed: 07/09/2012
• Issued: 01/28/2014
• Est. Priority Date: 10/24/2008
• Status: Active Grant

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Kawaguchi, Tomohiro
• Primary Examiner(s): LE, DIEU MINH T
• Application Number:
  US13/544,417
• Publication Number:
  US 20130003211A1
• Time in Patent Office: 568 Days
• Field of Search: 714/2, 714/5.1, 714/5.11, 714/6.1, 714/6.11, 714/6.12, 714/6.24, 714 10- 13
• US Class Current: 714/6.1
• CPC Class Codes: G06F 11/1076 Parity data used in redundant arrays of independent storages, e.g. in RAID systems
• CPC Class Codes: G06F 11/1092 Rebuilding, e.g. when physically replacing a failing disk

Abstract

A storage system comprises a first storage device having a first plurality of hard disk drives and a first controller. The first controller stores data in the first plurality of hard disk drives by stripes. Each stripe includes M data and N parity data allocated to M+N hard disk drives of the first plurality of hard disk drives. A first hard disk drive includes data or parity data of both a first stripe of the stripes and a second stripe of the stripes, while a second hard disk drive includes data or parity data of only one of the first stripe or the second stripe. During data recovery involving failure of one of the first plurality of hard disk drives, the data in the failed hard disk drive is recovered for each stripe by calculation using data and parity data in other hard disk drives for each stripe.

Claims

1. A storage system comprising:

a plurality of drives, wherein each of the plurality of drives is configured to include a plurality of parcels; and

a controller configured to;

provide a virtual volume to a host computer; and

store data into the plurality of drives by stripes, wherein each stripe includes M data and N parity data where the M and N are integers, the N parity data is calculated based on the M data for each stripe, the M data and N parity data are stored in M+N parcels of the plurality of parcels respectively, and each of the M+N parcels belongs to different M+N drives respectively,

wherein the plurality of parcels include first M+N parcels including a first stripe and second M+N parcels including a second stripe,

wherein a first drive of the plurality of drives includes both one of the first M+N parcels and one of the second M+N parcels, and a second drive of the plurality of drives includes either one of the first M+N parcels or one of the second M+N parcels, and

wherein a plurality of pages are created on the M+N parcels, where each of the pages includes a plurality of stripes and each of the pages is allocated to the virtual volume in page unit according to a write command.

2. The storage system according to claim 1,

wherein the plurality of drives are connected to a plurality of disk interfaces, and

wherein each drive including one of the first M+N parcels is connected to a different disk interface of the plurality of disk interfaces respectively.

3. The storage system according to claim 1,

wherein the plurality of drives are connected to a plurality of disk interfaces, and

wherein a number of the parcels of the first M+N parcels belonging to the drive which is connected to one disk interface is at most N.

4. The storage system according to claim 1,

wherein the controller is configured to provide a plurality of virtual volumes, and

wherein a first chunk made up by the first M+N parcels is allocated to a first virtual volume of the plurality of virtual volumes, and pages in the first chunk are allocated to the first virtual volume.

5. The storage system according to claim 1,

wherein during data recovery from failure of one of the plurality of drives as a failed drive including a failed parcel of the first M+N parcels, data stored in the failed parcel is recovered from data and parity data stored in parcels of the first M+N drives belonging to non-failed drives of the plurality of drives.

6. The storage system according to claim 5,

wherein data in the failed parcel is restored to a parcel in a non-failed drive which does not include the first M+N parcels.

7. A storage management method for a storage system, wherein the storage system has a plurality of drives and a controller, the method comprising:

providing a virtual volume to a host computer;

dividing each of the plurality of drives into a plurality of parcels; and

storing data into the plurality of drives by stripes, wherein each stripe includes M data and N parity data where the M and N are integers, the N parity data is calculated based on the M data for each stripe, the M data and N parity data are stored in M+N parcels of the plurality of parcels respectively, and each of the M+N parcels belongs to different M+N drives respectively,

wherein the plurality of parcels include first M+N parcels including a first stripe and second M+N parcels including a second stripe,

wherein a first drive of the plurality of drives includes both one of the first M+N parcels and one of the second M+N parcels, and a second drive of the plurality of drives includes either one of the first M+N parcels or one of the second M+N parcels, and

wherein a plurality of pages are created on the M+N parcels, where each of the pages includes a plurality of stripes and each of the pages is allocated to the virtual volume in page unit.

8. The storage management method according to claim 7,

wherein the plurality of drives are connected to a plurality of disk interfaces, and

wherein each drive including one of the first M+N parcels is connected to a different disk interface of the plurality of disk interfaces respectively.

9. The storage management method according to claim 7,

wherein the plurality of drives are connected to a plurality of disk interfaces, and

wherein a number of the parcels of the first M+N parcels belonging to the drive which is connected to one disk interface is at most N.

10. The storage management method according to claim 7, further comprising:

providing a plurality of virtual volumes,

making up a first chunk by the first M+N parcels;

allocating the first chunk to a first virtual volume of the plurality of virtual volumes; and

allocating pages in the first chunk to the first virtual volume.

11. The storage management method according to claim 7, further comprising:

during data recovery from failure of one of the plurality of drives as a failed drive including a failed parcel of the first M+N parcels, recovering data stored in the failed parcel is from data and parity data stored in parcels of the first M+N drives belonging to non-failed drives of the plurality of drives.

12. The storage management method according to claim 11, further comprising:

restoring data in the failed parcel to a parcel in a non-failed drive which does not include the first M+N parcels.

13. A non-transitory computer-readable medium storing a plurality of instructions for controlling a data processor to perform storage management for a storage system, wherein the storage system has a plurality of drives and a controller, the instructions comprising:

instructions that cause the data processor to provide a virtual volume to a host computer;

instructions that cause the data processor to divide each of the plurality of drives into a plurality of parcels; and

instructions that cause the data processor to store data into the plurality of drives by stripes, wherein each stripe includes M data and N parity data where the M and N are integers, the N parity data is calculated based on the M data for each stripe, the M data and N parity data are stored in M+N parcels of the plurality of parcels respectively, and each of the M+N parcels belongs to different M+N drives respectively,

wherein the plurality of parcels include first M+N parcels including a first stripe and second M+N parcels including a second stripe,

wherein a first drive of the plurality of drives includes both one of the first M+N parcels and one of the second M+N parcels, and a second drive of the plurality of drives includes either one of the first M+N parcels or one of the second M+N parcels, and

wherein a plurality of pages are created on the M+N parcels, where each of the pages includes a plurality of stripes and each of the pages is allocated to the virtual volume in page unit.

14. The non-transitory computer-readable medium according to claim 13,

wherein the plurality of drives are connected to a plurality of disk interfaces, and

wherein a number of the parcels of the first M+N parcels belonging to the drive which is connected to one disk interface is at most N.

15. The non-transitory computer-readable medium according to claim 13, wherein the instructions further comprise:

instructions that cause the data processor to provide a plurality of virtual volumes;

instructions that cause the data processor to make up a first chunk by the first M+N parcels;

instructions that cause the data processor to allocate the first chunk to a first virtual volume of the plurality of virtual volumes; and

instructions that cause the data processor to allocate pages in the first chunk to the first virtual volume.