Selecting optimal repair strategy for mirrored files

• US 20080010507A1
• Filed: 09/05/2006
• Published: 01/10/2008
• Est. Priority Date: 05/30/2006
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Vingralek, Radek
• Granted Patent: US 7,565,575 B2
• Time in Patent Office: Days
• US Class Current: 714/7
• CPC Class Codes: G06F 11/1092 Rebuilding, e.g. when physically replacing a failing disk

Abstract

This document describes solutions to reduce the time of reduced data redundancy following transient disk failures that do not corrupt the disk. Beneficially, these solutions provide a way to estimate the most efficient repair strategy for the disk group, which helps to minimize the amount of time data in a disk group remains unprotected. Merely by way of example, a threshold value might specify a duration in which a disk failure should be considered transient, such that if the disk is repaired within that duration, only the stale extents on the disk need be recreated. If the disk cannot be repaired within that duration, the entire contents of the disk might be recreated on one or more other disks in the group.

Claims

1. A method of repairing a disk group comprising a plurality of disks, the method comprising:

determining a threshold value for a repair time of the disk group, wherein the threshold value represents a duration during which the hard disk failure should be assumed to be transient;

maintaining a set of metadata for each hard disk in the disk group, wherein each hard disk stores a set of data blocks, each data block comprising a set of data stored on the hard disk, and wherein the set of metadata for a hard disk comprises information about whether each of the data blocks on the hard disk is current;

identifying with a computer a first hard disk in the disk group that has become unavailable, wherein the first hard disk comprises a first set of data blocks;

marking as stale, in the set of metadata for the first hard disk, each data block on the first hard disk to which a write attempt is made while the offline disk is unavailable; and

repairing the disk group, wherein repairing the disk group comprises;

(i) re-creating each of the data blocks marked as stale in the metadata for the first hard disk, if the first hard disk becomes available before the duration specified by the threshold value has expired; and

(ii) re-creating the first set of data blocks on one or more additional hard disks in the disk group, if the first hard disk does not become available before the duration specified by the threshold value has expired.

2. A method as recited in claim 1, wherein determining the threshold value comprises:

estimating a mean repair time for a transient failure of a hard disk in the disk group; and

calculating the threshold value based on the estimated mean repair time.

3. A method as recited in claim 2, wherein calculating the threshold value comprises multiplying the estimated mean repair time by a value based on one or more factors selected from the group consisting of:

a probability that a failure of the first hard disk is transient;

an expected duration required to re-create each of the data blocks marked as stale;

an expected duration required to re-create the first set of data blocks on the one or more additional hard disks.

4. A method as recited by claim 3, wherein the threshold value is calculated from the following formula:

5. A method as recited by claim 1, wherein determining the threshold value comprises allowing a user to specify the threshold value.

6. A method as recited by claim 1, wherein identifying the first hard disk as unavailable comprises marking the first hard disk as offline.

7. A method as recited by claim 1, further comprising mirroring each of the data blocks in the first set of data blocks on one or more other hard disks in the disk group.

8. A method as recited by claim 7, wherein re-creating a data block comprises copying the data block from a mirrored copy on another hard disk in the disk group.

9. A method as recited by claim 1, wherein (ii) further comprises removing the first hard disk from the disk group.

10. A method as recited by claim 1, wherein (i) comprises:

marking the first hard disk online for write operations;

re-creating on the first hard disk the set of metadata for the first hard disk;

copying to the first hard disk, and from one or more other hard disks in the disk group, each of the data blocks marked as stale in the metadata for the first hard disk;

marking the first hard disk as online for read operations; and

marking as current, in the metadata for the first hard disk, all of the copied data blocks.

11. A system for repairing a disk group, the system comprising:

a processor; and

a computer readable medium comprising a set of instructions executable by the processor, the set of instructions comprising;

(a) instructions to determine a threshold value for a repair time of the disk group, wherein the threshold value represents a duration during which the hard disk failure should be assumed to be transient;

(b) instructions to maintain a set of metadata for each hard disk in the disk group, wherein each hard disk stores a set of data blocks, each data block comprising a set of data stored on the hard disk, and wherein the set of metadata for a hard disk comprises information about whether each of the data blocks on the hard disk is current;

(c) instructions to identify a first hard disk in the disk group that has become unavailable, wherein the first hard disk comprises a first set of data blocks;

(d) instructions to mark as stale, in the set of metadata for the first hard disk, each data block on the first hard disk to which a write attempt is made while the offline disk is unavailable; and

(e) instructions to repair the disk group, wherein the instructions to repair the disk group comprise;

(i) instructions to re-create each of the data blocks marked as stale in the metadata for the first hard disk, if the first hard disk becomes available before the duration specified by the threshold value has expired; and

(ii) instructions to re-create the first set of data blocks on one or more additional hard disks in the disk group, if the first hard disk does not become available before the duration specified by the threshold value has expired.

12. A system, comprising:

a processor;

a disk group comprising a plurality of hard disks, each of the plurality of hard disks being in communication with the processor; and

a computer readable medium comprising a set of instructions executable by the processor, the set of instructions comprising;

(a) instructions to determine a threshold value for a repair time of the disk group, wherein the threshold value represents a duration during which the hard disk failure should be assumed to be transient;

(b) instructions to maintain a set of metadata for each hard disk in the disk group, wherein each hard disk stores a set of data blocks, each data block comprising a set of data stored on the hard disk, and wherein the set of metadata for a hard disk comprises information about whether each of the data blocks on the hard disk is current;

(c) instructions to identify a first hard disk in the disk group that has become unavailable, wherein the first hard disk comprises a first set of data blocks;

(d) instructions to mark as stale, in the set of metadata for the first hard disk, each data block on the first hard disk to which a write attempt is made while the offline disk is unavailable; and

(e) instructions to repair the disk group, wherein the instructions to repair the disk group comprise;

(i) instructions to re-create each of the data blocks marked as stale in the metadata for the first hard disk, if the first hard disk becomes available before the duration specified by the threshold value has expired; and

(ii) instructions to re-create the first set of data blocks on one or more additional hard disks in the disk group, if the first hard disk does not become available before the duration specified by the threshold value has expired.

13. A system as recited by claim 12, wherein the disk group comprises a redundant array of independent disks (“

RAID”

).

14. A system as recited by claim 12, wherein the set of metadata about each hard disk in the disk group is distributed across two or more of the hard disks, allowing the set of metadata for the first hard disk to be re-created from information on one or more other hard disks.

15. A computer program, embodied on a computer readable medium, for repairing a disk group comprising a plurality of hard disks, the computer program comprising a set of instructions executable by one or more computers, the set of instructions comprising:

instructions to determine a threshold value for a repair time of the disk group, wherein the threshold value represents a duration during which the hard disk failure should be assumed to be transient;

instructions to maintain a set of metadata for each hard disk in the disk group, wherein each hard disk stores a set of data blocks, each data block comprising a set of data stored on the hard disk, and wherein the set of metadata for a hard disk comprises information about whether each of the data blocks on the hard disk is current;

instructions to identify a first hard disk in the disk group that has become unavailable, wherein the first hard disk comprises a;

first set of data blocks;

instructions to mark as stale, in the set of metadata for the first hard disk, each data block on the first hard disk to which a write attempt is made while the offline disk is unavailable; and

instructions to repair the disk group, wherein the instructions to repair the disk group comprise;

(i) instructions to re-create each of the data blocks marked as stale in the metadata for the first hard disk, if the first hard disk becomes available before the duration specified by the threshold value has expired; and

(ii) instructions to re-create the first set of data blocks on one or more additional hard disks in the disk group, if the first hard disk does not become available before the duration specified by the threshold value has expired.

16. A system for repairing a disk group comprising a plurality of hard disks, the system comprising:

means for determining a threshold value for a repair time of the disk group, wherein the threshold value represents a duration during which the hard disk failure should be assumed to be transient;

means for maintaining a set of metadata for each hard disk in the disk group, wherein each hard disk stores a set of data blocks, each data block comprising a set of data stored on the hard disk, and wherein the set of metadata for a hard disk comprises information about whether each of the data blocks on the hard disk is current;

means for identifying a first hard disk in the disk group that has become unavailable, wherein the particular hard disk comprises a first set of data blocks;

means for marking as stale, in the set of metadata for the first hard disk, each data block on the first hard disk to which a write attempt is made while the offline disk is unavailable; and

means for repairing the disk group, wherein the means for repairing the disk group comprises;

(i) means for re-creating each of the data blocks marked as stale in the metadata for the first hard disk, if the first hard disk becomes available before the duration specified by the threshold value has expired; and

(ii) means for re-creating the first set of data blocks on one or more additional hard disks in the disk group, if the first hard disk does not become available before the duration specified by the threshold value has expired.