System, method and computer readable medium for obtaining consistent read performance for a plurality of flash drives or raid groups using workload and capacity limits

• US 10,254,970 B1
• Filed: 06/30/2016
• Issued: 04/09/2019
• Est. Priority Date: 06/30/2016
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Inventors: Martin, Owen, Wang, Hui, Alshawabkeh, Malak, Sahin, Adnan, Don, Arieh, Liu, Xiaomei
• Primary Examiner(s): Gossage, Glenn
• Application Number:
  US15/198,772
• Time in Patent Office: 1,013 Days
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0611 in relation to response time
• CPC Class Codes: G06F 3/0647 Migration mechanisms
• CPC Class Codes: G06F 3/0653 Monitoring storage devices or systems
• CPC Class Codes: G06F 3/0679 Non-volatile semiconductor memory device, e.g. flash memory, one time programmable memory [OTP]
• CPC Class Codes: G06F 3/0689 Disk arrays, e.g. RAID, JBOD

Abstract

Techniques for obtaining consistent read performance are disclosed that may include: receiving measured read I/O (input/output) response times for flash storage devices; and determining, in accordance with a specified allowable variation, whether a first of the measured read I/O response times for a first of the flash storage devices is inconsistent with respect to other ones of the measured read I/O response times. Responsive to determining the first measured read I/O response time is inconsistent first processing may be performed that corrects or alleviates the inconsistency of the first measured read I/O response time. The first processing may include varying the first measured read I/O response time of the first flash storage device by enforcing, for the first flash storage device, a write I/O workload limit a read I/O workload limit and an idle capacity limit. Data portions may be ranked and selected for data movement based on read workload, write workload or idle capacity. The flash storage may include storage devices of the same type or technology, and the same capacity. Response times for RAID groups may also be measured.

Claims

1. A method of obtaining consistent read performance comprising:

receiving a plurality of measured read input/output (I/O) response times, each of the plurality of measured read I/O response times denoting a measured read I/O response time for a different one of a plurality of flash storage devices;

determining, in accordance with a specified allowable variation, whether a first of the plurality of measured read I/O response times for a first of the plurality of flash storage devices is inconsistent with respect to other ones of the plurality of measured read I/O response times; and

responsive to determining the first measured read I/O response time is inconsistent with respect to other ones of the plurality of measured read I/O response times, performing first processing that corrects or alleviates the inconsistency of the first measured read I/O response time, the first processing including varying the first measured read I/O response time of the first flash storage device by enforcing, for the first flash storage device, a write I/O workload limit, a read I/O workload limit, and an idle capacity limit.

2. The method of claim 1, wherein the first processing includes:

receiving workload information for the first flash storage device, the workload information including a measured write I/O workload, a measured read I/O workload, and an amount of idle capacity;

determining whether the workload information for the first flash storage device violates any of a first write I/O workload limit, a read I/O workload limit, and a first idle capacity limit; and

responsive to determining the workload information for the first flash storage device violates any of the first write I/O workload limit, the read I/O workload limit, and the first idle capacity limit, performing a data movement that corrects or alleviates a detected violation of any of the first write I/O workload limit, the read I/O workload limit, and the first idle capacity limit.

3. The method of claim 2, wherein the first flash storage device has a first storage capacity and is a first type of flash technology, the plurality of flash storage devices includes a second flash storage device of the first type of flash technology, the second flash storage device having a second write I/O workload limit and having a second storage capacity different from the first storage capacity, and a write I/O workload ratio denotes a relative ratio of write performance achieved with the first storage capacity to write performance achieved with the second storage capacity, and wherein the first write I/O workload limit has a value determined in accordance with the write I/O workload ratio and the second write I/O workload limit.

4. The method of claim 3, wherein a first idle capacity ratio is a ratio of the first storage capacity relative to the second storage capacity, and the first idle capacity limit has a value determined in accordance with the first idle capacity ratio and a second idle capacity limit of the second flash storage device.

5. The method of claim 3, wherein the read I/O workload limit is the same limit enforced for the plurality of flash storage devices, and the write I/O workload ratio is a ratio of the first write I/O workload limit relative to the second write I/O workload limit.

6. The method of claim 5, further comprising:

determining whether the measured read I/O workload for the first flash storage device exceeds, by at least a specified amount, the read I/O workload limit;

responsive to determining that the measured read I/O workload for the first flash storage device exceeds, by at least a specified amount, the read I/O workload limit, performing second processing comprising;

ranking first data portions stored on the first flash storage device in accordance with decreasing read I/O workloads of the first data portions; and

selecting, in accordance with the ranking, a data portion set including one or more of the first data portions for movement from the first flash storage device to a target flash storage device of the plurality of flash storage devices.

7. The method of claim 6, wherein the target flash storage device is underutilized with respect to read I/O workload and wherein the one or more of the first data portions of the data portion set selected for movement to the target flash storage device do not result in a modeled read I/O workload for the target flash storage device exceeding the read I/O workload limit.

8. The method of claim 6, wherein each data portion in the data portion set has a read I/O workload that is greater than any other data portion of the first data portions not included in the data portion set.

9. The method of claim 3, wherein the plurality of flash storage devices includes a first set of flash storage devices of the first type of flash technology, the first set including the first flash storage device having the first storage capacity and the second flash storage device having the second storage capacity, and the method further includes:

determining, in accordance with a specified allowable variation, whether measured write I/O workloads of the flash storage devices of the first set are consistent; and

responsive to determining that a measured write I/O workload of any flash storage device of the first set is inconsistent, performing a data movement that alleviates or corrects an inconsistent measured write I/O workload of the any flash storage device of the first set.

10. The method of claim 2, further comprising:

determining whether the measured write I/O workload of the first flash storage device exceeds the first write I/O workload limit; and

responsive to determining the measured write I/O workload of the first flash storage device exceeds the first write I/O workload limit, performing second processing comprising;

ranking first data portions stored on the first flash storage device in accordance with increasing write I/O workloads of the first data portions; and

selecting, in accordance with the ranking, a data portion set including one or more of the first data portions for movement from the first flash storage device to a target flash storage device of the plurality of flash storage devices.

11. The method of claim 10, wherein the target flash storage device is underutilized with respect to write I/O workload and wherein the one or more of the first data portions of the data portion set selected for movement to the target flash storage device do not result in a modeled write I/O workload for the target flash storage device exceeding a specified write I/O workload limit of the target flash storage device.

12. The method of claim 10, wherein each data portion in the data portion set has a write I/O workload that is greater than any other data portion of the first data portions not included in the data portion set.

13. The method of claim 2, further comprising:

determining whether the amount of idle capacity of the first flash storage device exceeds the first idle capacity limit; and

responsive to determining the amount of idle capacity of the first flash storage device exceeds the first idle capacity limit, performing second processing comprising;

selecting a first set of one or more data portions stored on the first flash storage device for movement from the first flash storage device to a target flash storage device of the plurality of flash storage devices, each data portion of the first set having an idle workload.

14. The method of claim 1, wherein the plurality of flash storage devices includes at least a first flash storage device of a first flash technology and a second flash storage device of a second flash technology that is different from the first flash technology.

15. A system comprising:

a processor; and

a memory comprising code stored thereon that, when executed, performs a method of obtaining consistent read performance comprising;

receiving a plurality of measured read input/output (I/O) response times, each of the plurality of measured read I/O response times denoting a measured read I/O response time for a different one of a plurality of RAID groups, each of the plurality of RAID groups including flash storage devices of a same technology and a same capacity;

determining, in accordance with a specified allowable variation, whether a first of the plurality of measured read I/O response times for a first of the plurality of RAID groups is inconsistent with respect to other ones of the plurality of measured read I/O response times; and

responsive to determining the first measured read I/O response time is inconsistent with respect to other ones of the plurality of measured read I/O response times, performing first processing that corrects or alleviates the inconsistency of the first measured read I/O response time, the first processing including varying the first measured read I/O response time of the first RAID group by enforcing, for the first RAID group, a write I/O workload limit, a read I/O workload limit, and an idle capacity limit.

16. A non-transitory computer readable medium comprising code stored thereon that, when executed, performs a method of obtaining consistent read performance comprising:

receiving a plurality of measured read input/output (I/O) response times, each of the plurality of measured read I/O response times denoting a measured read I/O response time for a different one of a plurality of flash storage devices;

determining, in accordance with a specified allowable variation, whether a first of the plurality of measured read I/O response times for a first of the plurality of flash storage devices is inconsistent with respect to other ones of the plurality of measured read I/O response times; and

responsive to determining the first measured read I/O response time is inconsistent with respect to other ones of the plurality of measured read I/O response times, performing first processing that corrects or alleviates the inconsistency of the first measured read I/O response time, the first processing including varying the first measured read I/O response time of the first flash storage device by enforcing, for the first flash storage device, a write I/O workload limit, a read I/O workload limit, and an idle capacity limit.

17. The non-transitory computer readable medium of claim 16, wherein the first processing includes:

receiving workload information for the first flash storage device, the workload information including a measured write I/O workload, a measured read I/O workload, and an amount of idle capacity;

determining whether the workload information for the first flash storage device violates any of a first write I/O workload limit, a read I/O workload limit, and a first idle capacity limit; and

responsive to determining the workload information for the first flash storage device violates any of the first write I/O workload limit, the read I/O workload limit, and the first idle capacity limit, performing a data movement that corrects or alleviates a detected violation of any of the first write I/O workload limit, the read I/O workload limit, and the first idle capacity limit.

18. The non-transitory computer readable medium of claim 17, wherein the first flash storage device has a first storage capacity and is a first type of flash technology, the plurality of flash storage devices includes a second flash storage device of the first type of flash technology, the second flash storage device having a second write I/O workload limit and having a second storage capacity different from the first storage capacity, and a write I/O workload ratio denotes a relative ratio of write performance achieved with the first storage capacity to write performance achieved with the second storage capacity, and wherein the first write I/O workload limit has a value determined in accordance with the write I/O workload ratio and the second write I/O workload limit.

19. The non-transitory computer readable medium of claim 18, wherein a first idle capacity ratio is a ratio of the first storage capacity relative to the second storage capacity, and the first idle capacity limit has a value determined in accordance with the first idle capacity ratio and a second idle capacity limit of the second flash storage device.

20. The non-transitory computer readable medium of claim 18, wherein the read I/O workload limit is the same limit enforced for the plurality of flash storage devices, and the write I/O workload ratio is a ratio of the first write I/O workload limit relative to the second write I/O workload limit.