Apparatus and method for distributing writes asymmetrically among drives

• US 8,688,906 B2
• Filed: 11/16/2009
• Issued: 04/01/2014
• Est. Priority Date: 11/16/2009
• Status: Active Grant

• Current Assignee: Lenovo PC International Limited (Lenovo Group Ltd.)
• Original Assignee: Lenovo Singapore Pte Limited (Lenovo Group Ltd.)
• Inventors: Cromer, Daryl, Frame, Donald R., Hobbet, Jeffrey R., Mettler, Michael Scott
• Primary Examiner(s): Chery, Mardochee
• Application Number:
  US12/619,389
• Publication Number:
  US 20110119443A1
• Time in Patent Office: 1,597 Days
• Field of Search: 711/114, 711/E12.001
• US Class Current: 711/114
• CPC Class Codes: G06F 11/004 Error avoidance G06F11/07 and subgroups take precedence
• CPC Class Codes: G06F 11/1076 Parity data used in redundant arrays of independent storages, e.g. in RAID systems
• CPC Class Codes: G06F 12/0246 in block erasable memory, e.g. flash memory
• CPC Class Codes: G06F 2211/1059 Parity-single bit-RAID5, i.e. RAID 5 implementations
• CPC Class Codes: G06F 2212/1036 Life time enhancement
• CPC Class Codes: G06F 2212/7208 Multiple device management, e.g. distributing data over multiple flash devices
• CPC Class Codes: G06F 2212/7211 Wear leveling

Abstract

An apparatus and method are disclosed for distributing writes asymmetrically. An asymmetric distribution module calculates an asymmetric distribution of writes for a plurality of drives. A write module writes data to each drive in accordance with the asymmetric distribution of writes.

Claims

1. An apparatus comprising:

an asymmetric distribution module calculating an asymmetric distribution of writes for a plurality of drives with disproportionately more short-lived writes for a target drive by calculating a write threshold for each drive of the plurality of drives proportional to the asymmetric distribution of writes such that a probability based on the write thresholds for the plurality of drives that the target drive fails before any other drive of the plurality of drives fails exceeds a probability threshold; and

a write module writing data to each drive in accordance with the asymmetric distribution of writes.

2. The apparatus of claim 1, wherein the plurality of drives is organized as a Redundant Array of Independent Disks (RAID) and parity stripes are asymmetrically distributed among the plurality of drives in accordance with the asymmetric distribution of writes and the data is written to the plurality of drives according to a RAID scheme.

3. The apparatus of claim 2, wherein the RAID scheme is selected from the group consisting of a RAID 5 scheme, a RAID 6 scheme, a RAID 50 scheme, and a RAID 60 scheme.

4. The apparatus of claim 1, the write module writing the data to each drive in accordance with the asymmetric distribution of writes by:

designating a first drive of the plurality of drives as the target drive;

writing data to the target drive;

determining that the write threshold for the first drive is exceeded; and

designating a second drive of the plurality of drives as the target drive.

5. The apparatus of claim 1, wherein each write threshold is a specified number of writes and the asymmetric distribution is calculated using a Mean Writes Between Failures (MWBF) estimate of times a data block may be written to before the data block fails, wherein a failure distribution F(w) for each drive is calculated as F(w)=1−

e<sup>−

λ

w</sup> where λ

is a MWBF for the drive and w is the specified number of writes to the drive.

6. The apparatus of claim 1, wherein each of the plurality of drives are Solid State Drives (SSD).

7. The apparatus of claim 1, the write module redundantly storing first data across at least two drives of the plurality of drives.

8. A computer readable storage medium storing a computer readable program on a tangible storage device, wherein the computer readable program when executed on a computer causes the computer to:

calculate an asymmetric distribution of writes for a plurality of drives with disproportionately more short-lived writes for a target drive by calculating a write threshold for each drive of the plurality of drives proportional to the asymmetric distribution of writes such that a probability based on the write thresholds for the plurality of drives that the target drive fails before any other drive of the plurality of drives fails exceeds a probability threshold; and

write data to each drive in accordance with the asymmetric distribution of writes.

9. The computer readable storage medium of claim 8, wherein the plurality of drives are organized as a RAID and parity stripes are asymmetrically distributed among the plurality of drives in accordance with the asymmetric distribution of writes and the data is written to the plurality of drives according to a RAID scheme.

10. The computer readable storage medium of claim 9, wherein the RAID scheme is selected from the group consisting of a RAID 5 scheme, a RAID 6 scheme, a RAID 50 scheme, and a RAID 60 scheme.

11. The computer readable storage medium of claim 8, wherein the computer readable program further causes the computer to write the data to each drive in accordance with the asymmetric distribution of writes by:

designating a first drive of the plurality of drives as the target drive;

writing data to the target drive;

determining that the write threshold for the first drive is exceeded; and

designating a second drive of the plurality of drives as the target drive.

12. The computer readable storage medium of claim 8, wherein each write threshold is a specified number of writes and the asymmetric distribution is calculated using a Mean Writes Between Failures (MWBF) estimate of times a data block may be written to before the data block fails, wherein a failure distribution F(w) for each drive is calculated as F(w)=1−

e<sup>−

λ

w</sup> where λ

is a MWBF for the drive and w is the specified number of writes to the drive.

13. The computer readable storage medium of claim 8, wherein each of the plurality of drives are SSD.

14. The computer readable storage medium of claim 8, wherein the computer readable program further causes the computer to redundantly store first data across at least two drives of the plurality of drives.

15. A method comprising:

calculating an asymmetric distribution of writes for a plurality of drives with disproportionately more short-lived writes for a target drive by calculating a write threshold for each drive of the plurality of drives proportional to the asymmetric distribution of writes such that a probability based on the write thresholds for the plurality of drives that the target drive fails before any other drive of the plurality of drives fails exceeds a probability threshold; and

writing data to each drive in accordance with the asymmetric distribution of writes.

16. The method of claim 15, wherein the plurality of drives is organized as a RAID and parity stripes are asymmetrically distributed among the plurality of drives in accordance with the asymmetric distribution of writes and the data is written to the plurality of drives according to a RAID scheme.

17. The method of claim 16, wherein the RAID scheme is selected from the group consisting of a RAID 5 scheme, a RAID 6 scheme, a RAID 50 scheme, and a RAID 60 scheme.

18. The method of claim 15, the method writing the data to each drive in accordance with the asymmetric distribution of writes by:

designating a first drive of the plurality of drives as the target drive;

writing data to the target drive;

determining that the write threshold for the first drive is exceeded; and

designating a second drive of the plurality of drives as the target drive.

19. The method of claim 15, wherein each write threshold is a specified number of writes and the asymmetric distribution is calculated using a Mean Writes Between Failures (MWBF) estimate of times a data block may be written to before the data block fails, wherein a failure distribution F(w) for each drive is calculated as F(w)=1−

e<sup>−

λ

w</sup> where λ

is a MWBF for the drive and w is the specified number of writes to the drive.

20. The method of claim 15, wherein each of the plurality of drives are SSD.