HIGH PERFORMANCE DISK ARRAY REBUILD

• US 20090228648A1
• Filed: 03/04/2008
• Published: 09/10/2009
• Est. Priority Date: 03/04/2008
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: WACK, ANDREW P.
• Granted Patent: US 7,970,994 B2
• Time in Patent Office: Days
• US Class Current: 711/114
• CPC Class Codes: G06F 11/1092 Rebuilding, e.g. when physically replacing a failing disk

Abstract

A method, information processing system, and computer program storage product for managing disk failures in a disk array. At least one disk in a first disk array is determined to have failed. The first disk array only accesses data by reading an entire stripe that is distributed across multiple disks in the first disk array. A number of spare disks are selected that is equal to a number of disks making up the first disk. A second disk array is created from the number of disks that has been selected. Data is acquired from the first disk array by reading at least one entire stripe from the first disk array. A corresponding stripe comprising the data from each respective stripe that has been acquired is written to the second disk array.

Claims

1. A method for managing disk failures in a disk array, the method comprising:

determining that at least one disk has failed in a first disk array, wherein the first disk array only accesses data by reading an entire stripe that is distributed across multiple disks in the first disk array;

selecting a number of spare disks that is equal to a number of disks making up the first disk array;

creating a second disk array from the number of disks that has been selected;

acquiring data from the first disk array by reading at least one entire stripe from the first disk array; and

writing a corresponding stripe comprising the data from each respective stripe that has been acquired to the second disk array.

2. The method of claim 1, further comprising:

receiving a data read request for a requested data set;

determining that the first disk array and the second disk array both store a respective valid copy of the requested data set;

identifying that one of the first disk array and second disk array has a higher priority than the other;

writing, in response to the determining that the first disk array and the second disk array both store a respective valid copy of the requested data set, a second data set to a respective stripe on the one of the first disk array and second disk array with the higher priority; and

reading, in response to receiving the data read request and while writing the second set, the requested data set from one of the first disk array and the second disk array that does not have the higher priority.

3. The method of claim 1, further comprising:

determining that data from each stripe in the first disk array has been stored in the corresponding stripe in the second disk array; and

returning all of the disks in the first disk array except for the one disk that has failed to a pool of spare disks.

4. The method of claim 1, further comprising:

determining that an I/O request to write a specified data set has been received;

writing a new stripe comprising the specified data set to a corresponding stripe in the second disk array; and

reading, during the writing the new stripe, a second stripe from the first disk array corresponding to another stripe to be repaired the second disk array.

5. The method of claim 4, further comprising:

storing data comprising the second stripe in a memory,

wherein the writing the new stripe comprises writing the data comprising the second stripe that is contained in the memory to the second disk array, and wherein the writing the new stripe is performed in response to at least one of;

determining that an idle cycle exists for the second disk array; and

determining that data is being at least one of read from and written to the first disk array.

6. The method of claim 5, further comprising:

determining that the second stripe on the second disk array stores a newer version of data than the data comprising the second stripe that is stored the memory; and

removing, in response to determining that the second stripe on the second disk array stores a newer version of data than the data comprising the second stripe that is stored the memory, the data stored in memory.

7. The method of claim 1, wherein the first disk array and the second disk array are part of a General Parallel File System.

8. The method of claim 1, further comprising:

receiving a data read request for a requested data set;

determining that the second disk array stores a valid and more recent version of the requested data set than is stored in the first disk array; and

reading, in response to the determining that the second disk array stores a more valid and more recent version of the requested data set, the requested data set from a stripe at the second disk array comprising the requested data set.

9. The method of claim 8, further comprising:

determining that a pending write request exists in a queue to store s second data set; and

writing, while reading the data that has been requested from the stripe at the second disk array, a second data to the first disk array.

10. An information processing system for managing disk failures in a disk array, the information processing system comprising:

a memory;

a processor communicatively coupled to the memory; and

a disk array controller communicatively coupled to at least a first storage disk array and a set of spare storage disks, wherein the disk array controller comprises a disk rebuild manager adapted to;

determining that at least one disk has failed in the first disk array, wherein the first disk array only accesses data by reading an entire stripe that is distributed across multiple disks in the first disk array;

select a number of spare disks in the set of spare disks that is equal to a number of disks making up the first disk;

create a second disk array from the number of disks that has been selected;

acquire data from the first disk array by reading at least one entire stripe from the first disk array; and

write a corresponding stripe comprising the data from each respective stripe that has been acquired to the second disk array.

11. The information processing system of claim 10, wherein the disk array controller is further adapted to:

receive a data read request for a requested data set;

determine that the second disk array stores a valid and more recent version of the requested data set than is stored in the first disk array;

write, in response to the determining that the first disk array and the second disk array both store a respective valid copy of the requested data set, a second data set to a respective stripe on the one of the first disk array and second disk array with the higher priority; and

read, in response to the determining that the second disk array stores a more valid and more recent version of the requested data set, the requested data set from a stripe at the second disk array comprising the requested data set.

12. The information processing system of claim 11, wherein the disk array controller is further adapted to:

determine that a pending write request exists in a queue to store s second data set; and

write, while reading the data that has been requested from the stripe at the second disk array, a second data to the first disk array.

13. The information processing system of claim 10, wherein the disk array controller is further adapted to:

determine that data from each stripe in the first disk array has been stored in the corresponding stripe in the second disk array; and

return all of the disks in the first disk array except for the one disk that has failed to a pool of spare disks.

14. The information processing system of claim 10, wherein the disk array controller is further adapted to:

determine that an I/O request to write a specified data set has been received;

write a new stripe comprising the specified data set to a corresponding stripe in the second disk array; and

read, while the new stripe is being written, a second stripe from the first disk array corresponding to another stripe to be repaired the second disk array.

15. The information processing system of claim 14, wherein the disk array controller is further adapted to:

store data comprising the second stripe in a memory, wherein the disk array controller is further adapted to write the new stripe by;

writing the new stripe comprises writing the data comprising the second stripe that is contained in the memory to the second disk array, and wherein the writing the new stripe is performed in response to at least one of;

determining that an idle cycle exists for the second disk array; and

determining that data is being at least one of read from and written to the first disk array.

16. A computer program storage product for managing disk failures in a disk array, the computer program storage product tangibly encoded with instructions to implement a method, the method comprising:

determining that at least one disk has failed in a first disk array, wherein the first disk array only accesses data by reading an entire stripe that is distributed across multiple disks in the first disk array;

selecting a number of spare disks that is equal to a number of disks making up the first disk;

creating a second disk array from the number of disks that has been selected;

acquiring data from the first disk array by reading at least one entire stripe from the first disk array; and

writing a corresponding stripe comprising the data from each respective stripe that has been acquired to the second disk array.

17. The computer program storage product of claim 16, the method further comprising:

receiving a data read request for a requested data set;

determining that the second disk array stores a valid and more recent version of the requested data set than is stored in the first disk array; and

reading, in response to the determining that the second disk array stores a more valid and more recent version of the requested data set, the requested data set from a stripe at the second disk array comprising the requested data set.

18. The computer program storage product of claim 17, the method further comprising:

determining that a pending write request exists in a queue to store s second data set; and

writing, while reading the requested data set from the stripe at the second disk array, a second data to the first disk array.

19. The computer program storage product of claim 16, the method further comprising:

determining that an I/O request to write a specified data set has been received;

writing a new stripe comprising the specified data set to a corresponding stripe in the second disk array; and

reading, during the writing the new stripe, a second stripe from the first disk array corresponding to another stripe to be repaired the second disk array.

20. The computer program storage product of claim 19, the method further comprising:

storing data comprising the second stripe in a memory,

wherein the writing the new stripe comprises writing the data comprising the second stripe that is contained in the memory to the second disk array, and wherein the writing the new stripe is performed in response to at least one of;

determining that an idle cycle exists for the second disk array; and

determining that data is being at least one of read from and written to the first disk array.