Shifting wearout of storage disks

• US 10,528,276 B2
• Filed: 10/22/2015
• Issued: 01/07/2020
• Est. Priority Date: 10/22/2015
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Franke, Jeffery M., O'Connor, James A.
• Primary Examiner(s): Lo, Ann J
• Assistant Examiner(s): Loonan, Eric T
• Application Number:
  US14/920,129
• Publication Number:
  US 20170115897A1
• Time in Patent Office: 1,538 Days
• CPC Class Codes: G06F 11/1076 Parity data used in redundant arrays of independent storages, e.g. in RAID systems
• CPC Class Codes: G06F 2211/1073 Problems due to wear-out failures in RAID systems
• CPC Class Codes: G06F 3/0616 in relation to life time, e.g. increasing Mean Time Between Failures [MTBF]
• CPC Class Codes: G06F 3/0629 Configuration or reconfiguration of storage systems

Abstract

Technical solutions are described that shift wearout of an array of storage disks in a storage system. In an aspect, a method includes receiving a request to write data to the storage system. The method also includes generating checksum data corresponding to the data and dividing the checksum data into portions of distinct sizes. The method also includes writing each portion of the checksum data onto a respective storage disk in the array.

Claims

1. A storage system comprising:

a plurality of storage disks in an array; and

a controller that stores data across the storage disks, the controller being configured to;

generate a schedule to stagger wearout of the each storage disk in the array of storage disks at different times based on a mean time to failure (MTTF) of the storage disks in the array by;

determining a parity to write ratio of a parity to data according to a number of the storage disks,

calculating an average writes per week to reach a maximum write of at least one of the storage disks,

calculating a first write ratio for the at least one of the storage disks to reach the wearout an order of deviation earlier than the MTTF for the at least one of the storage disks, and

calculating a second write ratio for the at least one of the storage disks to reach wearout an order of deviation past the MTTF for the at least one of the storage disks;

associate with each storage disk in the array a corresponding predetermined offset for wearout with respect to the MTTF to schedule wearout of each storage disk at the corresponding predetermined offset from the MTTF, each corresponding predetermined offset being different for the each storage disk and based on the schedule to stagger the wearout of the storage disks at different times;

determine a write ratio for each storage disk in the array, the write ratio indicative of a number of writes to cause the storage disks to wearout according to the corresponding predetermined offsets of the respective storage disks; and

stagger, based at least in part on the schedule to stagger, a number of writes to the storage disks so that the storage disks wear out at the different times according to the each corresponding predetermined offset of the storage disks, the staggering comprising;

generating checksum data corresponding to the data to be written;

dividing the checksum data into a plurality of portions, a portion for each storage disk in the array, and each portion of a different size, wherein the different sizes correspond to the write ratios for each storage; and

writing each portion of the checksum data onto a respective storage disk in the array.

2. The system of claim 1, wherein the sizes of the portions of the checksum data are based on the schedule to stagger the wearout of the storage disks.

3. The system of claim 2, wherein a portion of the checksum data is assigned to the respective storage disk in the array based on an inverse relationship between the size of the portion and the mean time to failure of said storage disk.

4. The system of claim 2, wherein the schedule to stagger the wearout of the storage disks staggers consecutive wearouts of storage disks in the array by the corresponding predetermined offset that represents a predetermined duration, wherein the predetermined duration represents an average time to replace and rebuild a worn out storage disk in the array.

5. The system of claim 2, wherein a storage disk in the array is one of a solid state drive, a flash memory, or a flash dual in-line memory module (DIMM).

6. The system of claim 5, wherein the array of storage disks is a redundant array of independent disks (RAID).

7. The system of claim 1, wherein a maximum predetermined offset is determined as (N/2)*2σ

if (N) is an odd number and the maximum predetermined offset is determined as (N/2*2σ

)−

σ

if (N) is an even number, where (N) is a number of the storage disks in the array and (σ

) is a predetermined value, wherein at least one of the first write ratio or the second write ratio is calculated based at least in part on the maximum predetermined offset.

8. A computer program product for staggering wearout of storage disks arranged in an array, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:

generate a schedule to stagger wearout of the each storage disk in the array of storage disks at different times based on a mean time to failure (MTTF) of the storage disks in the array by;

determining a parity to write ratio of a parity to data according to a number of the storage disks,

calculating an average writes per week to reach a maximum write of at least one of the storage disks,

calculating a first write ratio for the at least one of the storage disks to reach the wearout an order of deviation earlier than the MTTF for the at least one of the storage disks, and

calculating a second write ratio for the at least one of the storage disks to reach wearout an order of deviation past the MTTF for the first storage disk;

associate with each storage disk in the array a corresponding predetermined offset for wearout with respect to the MTTF to schedule wearout of each storage disk at the corresponding predetermined offset from the MTTF, each corresponding predetermined offset being different for the each storage disk and based on the schedule to stagger the wearout of the storage disks at different times;

determine a write ratio for each storage disk in the array, the write ratio indicative of a number of writes to cause the storage disks to wearout according to the corresponding predetermined offsets of the respective storage disks; and

stagger, based at least in part on the schedule to stagger, a number of writes to the storage disks so that the storage disk wears out at the different times according to the each corresponding predetermined offset of the storage disks, the staggering comprising;

generation of checksum data corresponding to the data to be written;

division of the checksum data into a plurality of portions, a portion for each storage disk in the array, and each portion of, a different size, wherein the sizes of the portions of the checksum data are based on the write ratios for each storage disk; and

writing of each portion of the checksum data onto a respective storage disk in the array.

9. The computer product of claim 8, wherein a portion of the checksum data is assigned to the respective storage disk in the array based on an inverse relationship between the size of the portion and the mean time to failure of said storage disk.

10. The computer product of claim 8, wherein the schedule to stagger the wearout of the storage disks staggers consecutive wearouts of storage disks in the array by a predetermined duration prior, wherein the predetermined duration represents an average time to replace and rebuild a worn out storage disk in the array.

11. The computer program product of claim 8, wherein a maximum predetermined offset is determined as (N/2)*2σ

if (N) is an odd number and the maximum predetermined offset is determined as (N/2*2σ

)−

σ

if (N) is an even number, where (N) is a number of the storage disks in the array and (σ

) is a predetermined value, wherein at least one of the first write ratio or the second write ratio is calculated based at least in part on the maximum predetermined offset.