Location of data among storage tiers

• US 9,323,655 B1
• Filed: 06/30/2011
• Issued: 04/26/2016
• Est. Priority Date: 06/30/2011
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Sahin, Adnan, Veprinsky, Alexandr, Marshak, Marik, Wang, Hui, Liu, Xiaomei, Martin, Owen, Dolan, Sean C.
• Primary Examiner(s): Tsai, Sheng-Jen
• Application Number:
  US13/135,265
• Time in Patent Office: 1,762 Days
• Field of Search: 711/173
• US Class Current: 1/1
• CPC Class Codes: G06F 12/023 Free address space management
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0617 in relation to availability
• CPC Class Codes: G06F 3/0649 Lifecycle management
• CPC Class Codes: G06F 3/0665 at area level, e.g. provisioning of virtual or logical volumes
• CPC Class Codes: G06F 3/0683 Plurality of storage devices
• CPC Class Codes: G06F 3/0685 Hybrid storage combining heterogeneous device types, e.g. hierarchical storage, hybrid arrays

Abstract

Managing data in a storage system having a plurality of classes of storage includes determining an amount of data to be provided on at least one of the classes of storage according to a policy, dynamically setting a threshold according to the amount of data to be provided on the at least one of the classes or an expected performance based on the threshold, and placing data on particular classes of storage based on the threshold. Dynamically setting a threshold may include sorting data portions according to at least one score thereof and may include determining a particular score corresponding to the amount of data to be provided. The data portions may be provided in a histogram having a horizontal scale corresponding to a score value and a vertical scale corresponding to a number of data portions having a particular value.

Claims

1. A method of managing data in a storage system having a plurality of classes of storage, comprising:

determining an amount of data to be provided on at least one of the classes of storage according to a policy;

generating a score for each data portion in a plurality of data portions in the data based on activity for a respective data portion;

dynamically setting a threshold for providing data to the at least one of the classes of storage according to one or more of;

the an amount of data to be provided on the at least one of the classes of storage or an expected performance based on the threshold, wherein dynamically setting the threshold according to the amount of data to be provided on the at least one of the classes of storage includes determining a score range corresponding to the amount of data to be provided based on determining a collective amount of data in data portions within the score range, with an end of the score range being used to set the threshold, and wherein dynamically setting the threshold according to the expected performance based on the threshold includes limiting a value of the threshold to avoid placing data portions with a score on one side of the threshold on the at least one of the classes of storage; and

placing data on particular classes of storage based on the threshold.

2. A method, according to claim 1, wherein the series of categories form a histogram having a horizontal scale corresponding to a score value and a vertical scale corresponding to a number of data portions having a particular value.

3. A method, according to claim 1, wherein at least some of the scores are logarithmic.

4. A method, according to claim 1, wherein the data portions are extents having five thousand seven hundred and sixty tracks of data.

5. A method, according to claim 1, wherein the data portions are extents, and wherein contiguous extents having scores that are the same are combined into superextents.

6. A method, according to claim 1, wherein at least one score is based on access statistics, and wherein the access statistics include read misses, writes, and prefetches.

7. A method, according to claim 1, wherein placing data includes promotion of some of the data to relatively faster storage classes having a score above a first threshold and demotion of some of the data to relatively slower storage classes having a score below a second threshold.

8. A method, according to claim 7, wherein a data portion that is marked for promoting based on a score thereof is not considered for demotion for a time interval according to the policy.

9. A method, according to claim 7, wherein there are three storage classes of data and wherein data that is marked for promoting to a fastest class of storage is not considered for demotion and wherein data that is marked for promoting to a second fastest class of storage is not considered for demotion to a slowest class of storage.

20. A method according to claim 1, wherein at least some of the scores are based on a non-linear scale.

10. Computer software, provided in a non-transitory computer-readable medium, that manages data in a storage system having a plurality of classes of storage, the software comprising:

executable code that determines an amount of data to be provided on at least one of the classes of storage according to a policy;

executable code that generates a score for each data portion in a plurality of data portions in the data based on activity for a respective data portion;

executable code that dynamically sets a threshold for providing data to the at least one of the classes of storage according to one or more of;

the amount of data to be provided on the at least one of the classes of storage or an expected performance based on the threshold, wherein the executable code that dynamically sets the threshold according to the amount of data to be provided on the at least one of the classes of storage includes executable code that determines a score range corresponding to the amount of data to be provided based on determining a collective amount of data in data portions within the score range, with an end of the score range being used to set the threshold, and wherein the executable code that dynamically setting the threshold according to the expected performance based on the threshold includes executable code that limits a value of the threshold to avoid placing data portions with a score on one side of the threshold on the at least one of the classes of storage; and

executable code that places data on particular classes of storage based on the threshold.

11. Computer software, according to claim 10, wherein the series of categories form a histogram having a horizontal scale corresponding to a score value and a vertical scale corresponding to a number of data portions having a particular value.

12. Computer software, according to claim 10, wherein at least some of the scores are logarithmic.

13. Computer software, according to claim 10, wherein the data portions are extents having five thousand seven hundred and sixty tracks of data.

14. Computer software, according to claim 10, wherein the data portions are extents, and wherein contiguous extents having scores that are the same are combined into superextents.

15. Computer software, according to claim 10, wherein at least one score is based on access statistics, and wherein the access statistics include read misses, writes, and prefetches.

16. Computer software, according to claim 10, wherein executable code that places data includes executable code that promotes some of the data to relatively faster storage classes having a score above a first threshold and demotes some of the data to relatively slower storage classes having a score below a second threshold.

17. Computer software, according to claim 16, wherein a data portion that is marked for promoting based on a score thereof is not considered for demotion for a time interval according to the policy.

18. Computer software, according to claim 16, wherein there are three storage classes of data and wherein data that is marked for promoting to a fastest class of storage is not considered for demotion and wherein data that is marked for promoting to a second fastest class of storage is not considered for demotion to a slowest class of storage.

19. Computer software, according to claim 10, wherein at least some of the scores are based on a non-linear scale.