Selective defragmentation based on IO hot spots

• US 8,909,887 B1
• Filed: 09/25/2012
• Issued: 12/09/2014
• Est. Priority Date: 09/25/2012
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Armangau, Philippe, Bono, Jean-Pierre, Seibel, Chris, Chen, Xiangping
• Primary Examiner(s): Blair, April Y
• Assistant Examiner(s): Cygiel, Gary W
• Application Number:
  US13/626,422
• Time in Patent Office: 805 Days
• Field of Search: 711/165, 707/693
• US Class Current: 711/165
• CPC Class Codes: G06F 12/0866 for peripheral storage systems, e.g. disk cache
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0643 Management of files
• CPC Class Codes: G06F 3/0644 Management of space entities, e.g. partitions, extents, pools
• 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

Abstract

A technique of managing data fragmentation is performed in a data storage apparatus. The technique involves receiving an indication that IO access to particular data stored within a memory space has exceeded a predefined threshold level. The technique further involves performing, in response to the indication, a defragmentation evaluation operation. The technique further involves performing a set of data movement operations which moves the particular data from a first set of storage locations of the memory space to a second set of storage locations of the memory space to improve defragmentation of the particular data when a result of the defragmentation evaluation operation has a first value. The technique further involves maintaining the particular data at the first set of storage locations of the memory space when the result of the defragmentation evaluation operation has a second value which is different than the first value.

Claims

1. In a data storage apparatus, a method of managing data fragmentation, the method comprising:

receiving an indication that input/output (IO) access to particular data stored within a memory space has exceeded a predefined threshold level;

in response to the indication that IO access to the particular data stored within the memory space has exceeded the predefined threshold level, performing a defragmentation evaluation operation;

when a result of the defragmentation evaluation operation has a first value, performing a set of data movement operations which moves the particular data from a first set of storage locations of the memory space to a second set of storage locations of the memory space to improve defragmentation of the particular data; and

when the result of the defragmentation evaluation operation has a second value which is different than the first value, maintaining the particular data at the first set of storage locations of the memory space;

wherein receiving the indication that IO access to the particular data stored within the memory space has exceeded the predefined threshold level includes;

processing a historical log of locks imposed on the memory space to ascertain the number of times the particular data stored within the memory space was accessed during a recent time period;

wherein the historical log of locks imposed on the memory identifies ranges of extents which were locked during IO operations performed on the memory space; and

wherein processing the historical log of locks imposed on the memory space includes updating a set of counters based on the ranges of extents which were locked during the IO operations and which are identified by the historical log of locks.

2. A method as in claim 1 wherein the memory space includes multiple tiers of non-volatile storage which are tiered based on IO access time;

and

wherein performing the set of data movement operations which moves the particular data from the first set of storage locations of the memory space to the second set of storage locations of the memory space includes transferring the particular data from an initial tier of non-volatile storage to a faster tier of non-volatile storage, the faster tier of non-volatile storage providing faster IO access time than the initial tier of non-volatile storage.

3. A method as in claim 1 wherein the memory space includes multiple tiers of non-volatile storage which are tiered based on IO access time;

and

wherein performing the set of data movement operations which moves the particular data from the first set of storage locations of the memory space to the second set of storage locations of the memory space includes transferring the particular data among storage locations of the same tier of non-volatile storage to prevent a decrease in IO access time.

4. A method as in claim 1 wherein the memory space is partitioned into logical unit (LUN) slices;

and

wherein performing the set of data movement operations which moves the particular data from the first set of storage locations of the memory space to the second set of storage locations of the memory space includes transferring the particular data to a contiguous sequence of storage locations of a single LUN slice.

5. A method as in claim 4 wherein the data storage apparatus maintains a pool of empty LUN slices;

and

wherein transferring the particular data to the contiguous sequence of storage locations of the single LUN slice includes (i) allocating a LUN slice from the pool of empty LUN slices and (ii) storing the particular data in that allocated LUN slice.

6. A method as in claim 4 wherein the data storage apparatus maintains partially filled LUN slices;

and

wherein transferring the particular data to the contiguous sequence of storage locations of the single LUN slice includes (i) identifying a particular LUN slice of the partially filled LUN slices and (ii) storing the particular data in an empty portion of that particular LUN slice.

7. A method as in claim 1 wherein a subset of the counters identifies a level of IO access to the particular data;

and wherein the method further comprises;

after performing the set of data movement operations, clearing the subset of the counters.

8. A method as in claim 1, further comprising:

after performing the set of data movement operations, creating a defrag timestamp record to indicate when the particular data stored within the memory space was last defragmented.

9. A method as in claim 1 wherein updating the set of counters includes:

updating the set of counters in response to shared locks imposed on ranges of blocks during host read operations.

10. A method as in claim 1 wherein updating the set of counters includes:

updating the set of counters in response to exclusive locks imposed on ranges of blocks during host write operations.

11. A data storage apparatus which manages data fragmentation, the data storage apparatus comprising:

a host interface;

memory which provides a memory space; and

processing circuitry coupled to the host interface and the memory, the processing circuitry being constructed and arranged to;

receive an indication that input/output (IO) access to particular data stored within the memory space has exceeded a predefined threshold level,

in response to the indication that IO access to the particular data stored within the memory space has exceeded the predefined threshold level, perform a defragmentation evaluation operation,

when a result of the defragmentation evaluation operation has a first value, perform a set of data movement operations which moves the particular data from a first set of storage locations of the memory space to a second set of storage locations of the memory space to improve defragmentation of the particular data, and

when the result of the defragmentation evaluation operation has a second value which is different than the first value, maintain the particular data at the first set of storage locations of the memory space;

wherein the processing circuitry, when receiving the indication that IO access to the particular data stored within the memory space has exceeded the predefined threshold level, is constructed and arranged to;

process a historical log of locks imposed on the memory space to ascertain the number of times the particular data stored within the memory space was accessed during a recent time period;

wherein the historical log of locks imposed on the memory identifies ranges of extents which were locked during IO operations performed on the memory space; and

wherein the processing circuitry, when processing the historical log of locks imposed on the memory space, is constructed and arranged to update a set of counters based on the ranges of extents which were locked during the IO operations and which are identified by the historical log of locks.

12. A data storage apparatus as in claim 11 wherein the memory includes multiple tiers of non-volatile storage which are tiered based on IO access time;

and

wherein the processing circuitry, when performing the set of data movement operations which moves the particular data from the first set of storage locations of the memory space to the second set of storage locations of the memory space, is constructed and arranged to transfer the particular data from an initial tier of non-volatile storage to a faster tier of non-volatile storage, the faster tier of non-volatile storage providing faster IO access time than the initial tier of non-volatile storage.

13. A data storage apparatus as in claim 11 wherein the memory includes multiple tiers of non-volatile storage which are tiered based on IO access time;

and

wherein the processing circuitry, when performing the set of data movement operations which moves the particular data from the first set of storage locations of the memory space to the second set of storage locations of the memory space, is constructed and arranged to transfer the particular data among storage locations of the same tier of non-volatile storage to prevent a decrease in IO access time.

14. A data storage apparatus as in claim 11 wherein the memory space is partitioned into logical unit (LUN) slices;

and

wherein the processing circuitry, when performing the set of data movement operations which moves the particular data from the first set of storage locations of the memory space to the second set of storage locations of the memory space, is constructed and arranged to transfer the particular data to a contiguous sequence of storage locations of a single LUN slice.

15. A data storage apparatus as in claim 14 wherein the memory space includes a pool of empty LUN slices;

and

wherein the processing circuitry, when transferring the particular data to the contiguous sequence of storage locations of the single LUN slice, is constructed and arranged to (i) allocate a LUN slice from the pool of empty LUN slices and (ii) store the particular data in that allocated LUN slice.

16. A data storage apparatus as in claim 14 wherein the memory space includes partially filled LUN slices;

and

wherein the processing circuitry, when transferring the particular data to the contiguous sequence of storage locations of the single LUN slice, is constructed and arranged to (i) identify a particular LUN slice of the partially filled LUN slices and (ii) store the particular data in an empty portion of that particular LUN slice.

17. A data storage apparatus as in claim 11 wherein the processing circuitry, when updating the set of counters, is constructed and arranged to:

update the set of counters in response to shared locks imposed on ranges of blocks during host read operations.

18. A data storage apparatus as in claim 11 wherein the processing circuitry, when updating the set of counters, is constructed and arranged to:

update the set of counters in response to exclusive locks imposed on ranges of blocks during host write operations.

19. A computer program product having a non-transitory computer readable medium storing a set of instructions which, when carried out by computerized circuitry, directs the computerized circuitry to manage data fragmentation by performing a method comprising:

receiving an indication that input/output (IO) access to particular data stored within a memory space has exceeded a predefined threshold level;

in response to the indication that IO access to the particular data stored within the memory space has exceeded the predefined threshold level, performing a defragmentation evaluation operation;

when a result of the defragmentation evaluation operation has a first value, performing a set of data movement operations which moves the particular data from a first set of storage locations of the memory space to a second set of storage locations of the memory space to improve defragmentation of the particular data; and

when the result of the defragmentation evaluation operation has a second value which is different than the first value, maintaining the particular data at the first set of storage locations of the memory space;

wherein receiving the indication that IO access to the particular data stored within the memory space has exceeded the predefined threshold level includes;

processing a historical log of locks imposed on the memory space to ascertain the number of times the particular data stored within the memory space was accessed during a recent time period; and

wherein processing the historical log of locks imposed on the memory space includes updating a set of counters based on the ranges of extents which were locked during the IO operations and which are identified by the historical log of locks.

20. A computer program product as in claim 19 wherein updating the set of counters includes:

updating the set of counters in response to shared locks imposed on ranges of blocks during host read operations.

21. A computer program product as in claim 19 wherein updating the set of counters includes:

updating the set of counters in response to exclusive locks imposed on ranges of blocks during host write operations.