Techniques for statistics collection in connection with data storage performance

• US 8,583,838 B1
• Filed: 09/27/2010
• Issued: 11/12/2013
• Est. Priority Date: 09/27/2010
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Marshak, Marik, Martin, Owen, Veprinsky, Alex, Naamad, Amnon, Sahin, Adnan, Dolan, Sean C.
• Primary Examiner(s): Alrobaye, Idriss N
• Assistant Examiner(s): Huson, Zachary K
• Application Number:
  US12/924,396
• Time in Patent Office: 1,142 Days
• Field of Search: None
• US Class Current: 710/18
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0619 in relation to data integrity, e.g. data losses, bit errors
• CPC Class Codes: G06F 3/0647 Migration mechanisms
• CPC Class Codes: G06F 3/0649 Lifecycle management
• CPC Class Codes: G06F 3/0685 Hybrid storage combining heterogeneous device types, e.g. hierarchical storage, hybrid arrays
• CPC Class Codes: G06F 3/0689 Disk arrays, e.g. RAID, JBOD

Abstract

Described are techniques for tracking activity for one or more devices. Each of the one or more devices is partitioned into one or more extents. Each of the one or more extents includes a plurality of sub extents. Each of the sub extents includes a plurality of chunks. For each of said one or more extents of each of said one or more devices, extent activity level information is determined including one or more metrics indicating an activity level for said each extent; and For each of said one or more extents of each of said one or more devices, an activity bitmap is determined. The activity bitmap includes an entry for each sub extent included in said each extent, said entry indicating an activity level for said each sub extent.

Claims

1. A method for tracking and using activity information for one or more logical devices comprising:

partitioning each of said one or more logical devices into one or more extents, each of said one or more extents including a plurality of sub extents, each of said sub extents including a plurality of chunks;

determining, for each of said one or more extents of each of said one or more logical devices, extent activity level information including one or more metrics indicating an activity level for said each extent;

determining, for each of said one or more extents of each of said one or more logical devices, an activity bitmap, said activity bitmap including an entry for each sub extent included in said each extent, said entry indicating an activity level for said each sub extent; and

selecting a first sub extent of a first extent of a first of the one or more logical devices as a candidate for data movement using said extent activity level information and said activity bitmap for said first extent, said data movement being any of demotion from a first storage tier to a second storage tier whereby the first storage tier is a higher performance tier than said second storage tier, and promotion from the second storage tier to the first storage tier, said first storage tier including physical storage devices of a first storage type and said second storage tier including physical storage devices of a second storage type different than the first storage type, wherein a physical storage device of the first storage type provides higher performance than a physical storage device of the second storage type.

2. The method of claim 1, wherein each of said one or more logical devices is a thin device having a storage capacity wherein storage is unallocated for at least a first portion of said storage capacity at a point in time, said first portion corresponding to a range of addresses of said thin device not mapped to physical storage.

3. The method of claim 1, wherein said extent activity level information includes short term activity information and long term activity information.

4. The method of claim 3, wherein said short term activity information includes a first metric determined using a first decay coefficient providing a rate of decay for an activity level determined at a point in time, and said long term activity information includes a second metric determined using a second decay coefficient providing a rate of decay for an activity level determined at a point in time.

5. The method of claim 4, wherein said first decay coefficient specifies a shorter rate of decay than said second decay coefficient.

6. The method of claim 3, wherein said short term activity information and said long term activity information each include a read miss rate, a write rate, and a prefetch rate.

7. The method of claim 6, wherein said short term activity information and said long term activity information are determined based on a number of operations performed by a device interface which accesses physical devices.

8. The method of claim 1, wherein said first sub extent has data stored on the first storage tier and a second sub extent of the first extent has data stored on a second storage tier.

9. The method of claim 1, wherein the first sub extent is identified as a candidate for movement from the second storage tier to the first storage tier.

10. The method of claim 1, wherein the first sub extent is identified as a candidate for movement from the first storage tier to the second storage tier.

11. The method of claim 1, further comprising determining device activity level information for each of said one or more logical devices, wherein said device activity level information for each of said one or more logical devices is used to identify said first logical devices for data movement, said extent activity level information for each extent of said first logical device is used to identify said first extent for data movement, and said activity bitmap for said first extent is used to identify one or more sub extents, including the first sub extent, of said first extent for data movement.

12. The method of claim 2, wherein said logical devices have data stored on a plurality of physical devices included in a data storage system having a plurality of device interfaces, and for each of said device interfaces processing an I/O operation directed to a thin device, said each device interface writes an I/O trace record including thin device information for the I/O operation in a first buffer included in memory local to said each device interface.

13. The method of claim 12, wherein, when said first buffer is full, said each device interface performs processing to flush a plurality of I/O trace records from said first buffer and aggregate activity information of said device interface for one or more thin devices as determined in accordance with said plurality of I/O trace records.

14. The method of claim 13, wherein said each device interface updates a master copy of activity information in global memory in accordance with said activity information of said device interface, said master copy including activity information for said plurality of device interfaces.

15. The method of claim 2, wherein, at each sampling period for which activity information is updated for said one or more logical devices, each of said plurality of device interfaces performs processing for a portion of said one or more logical devices.

20. The method of claim 1, wherein physical devices of the second storage tier are solid state storage devices and physical devices of the first storage tier are non-solid state storage devices.

16. A data storage system comprising:

a plurality of physical devices;

a plurality of logical devices mapped to portions of said physical devices, wherein said plurality of logical devices includes one or more thin devices, each of said thin devices being partitioned into one or more extents, each of said extents being partitioned into one or more sub extents;

a plurality of device interfaces used to access said plurality of physical devices;

a global memory accessible to said plurality of device interfaces; and

wherein each of said plurality of device interfaces services I/Os for a portion of said physical devices, each of said device interfaces including a memory comprising code stored thereon for;

recording first information for each I/O serviced by said each device interface for one of said plurality of thin devices;

aggregating said first information for a plurality of I/Os serviced by said each device interface and producing local aggregation information;

updating a portion of said global memory including second information, said second information comprising local aggregation information for said plurality of device interfaces;

processing, at each sampling period, a portion of said second information corresponding to a portion of said plurality of thin devices to produce activity information;

determining, for each of said one or more extents of each of the plurality of logical devices, extent activity level information and an activity bitmap including an entry for each sub extent in said each extent; and

selecting a first sub extent of a first extent of a first of the plurality of logical devices as a candidate for data movement using said extent activity level information and said activity bitmap for said first extent, said data movement being any of demotion from a first storage tier to a second storage tier whereby the first storage tier is a higher performance tier than said second storage tier, and promotion from the second storage tier to the first storage tier, said first storage tier including physical storage devices of a first storage type and said second storage tier including physical storage devices of a second storage type different than the first storage type, wherein a physical storage device of the first storage type provides higher performance than a physical storage device of the second storage type.

17. The data storage system of claim 16, wherein said extent activity level information includes short term activity information and long term activity information, said short term activity information includes a first metric determined using a first decay coefficient providing a rate of decay for an activity level determined at a point in time, and said long term activity information includes a second metric determined using a second decay coefficient providing a rate of decay for an activity level determined at a point in time.

18. The data storage system of claim 16, wherein each of said sub extents corresponds to a portion of a physical device and represents a smallest amount considered as a data movement candidate, said data storage system including a plurality of storage tiers, each of said plurality of storage tiers having physical devices of different performance characteristics and including at least the first storage tier and the second storage tier wherein said first sub extent is included on one of said plurality of thin devices and is a data movement candidate having a first portion of data stored on a physical device of said first storage tier and a second portion of data stored on a physical device of said second storage tier and wherein, when said first sub extent is selected as a candidate for promotion to said first storage tier, said second portion is moved to a physical device of said first storage tier, and when said first sub extent is selected as a candidate for demotion to said second storage tier, said first portion is moved to a physical device of said second storage tier.

19. A non-transitory computer readable medium comprising code stored thereon for tracking and using activity information for one or more logical devices, the non-transitory computer readable medium comprising code stored thereon for:

partitioning each of said one or more logical devices into one or more extents, each of said one or more extents including a plurality of sub extents, each of said sub extents including a plurality of chunks;

determining, for each of said one or more extents of each of said one or more logical devices, extent activity level information including one or more metrics indicating an activity level for said each extent;

determining, for each of said one or more extents of each of said one or more logical devices, extent activity bitmap, said activity bitmap including an entry for each sub extent included in said each extent, said entry indicating an activity level for said each sub extent; and

selecting a first sub extent of a first extent of a first of the one or more logical devices as a candidate for data movement using said extent activity level information and said activity bitmap for said first extent, said data movement being any of demotion from a first storage tier to a second storage tier whereby the first storage tier is a higher performance tier than said second storage tier, and promotion from the second storage tier to the first storage tier, said first storage tier including physical storage devices of a first storage type and said second storage tier including physical storage devices of a second storage type different than the first storage type, wherein a physical storage device of the first storage type provides higher performance than a physical storage device of the second storage type.