Proactive release of high performance data storage resources when exceeding a service level objective

1. A method of proactively releasing high performance storage resources, comprising:

allocating, by a storage processor in a data storage system, units of storage from multiple physical disk groups to one or more storage objects in a storage group, wherein the storage objects are used by at least one application to store host data, wherein the multiple physical disk groups include a high performance disk group made up of one or more high performance disk drives and at least one lower performance disk group made up of one or more lower performance disk drives, wherein the storage group is associated with a service level objective, and wherein the service level objective comprises an average response time target for I/O operations performed on the storage objects in the storage group;

calculating a load score for each one of the units of storage allocated to the storage group from the high performance disk group;

determining a minimum load score equal to the lowest load score among the load scores calculated for the units of storage allocated from the high performance disk group to the storage group;

detecting that an average response time for the storage group is exceeding the service level objective associated with the storage group; and

in response to detecting that the average response time for the storage group is exceeding the service level objective associated with the storage group,

calculating a demotion threshold based on the minimum load score, wherein the demotion threshold is a product of multiplying the minimum load score and a demotion range factor,

identifying a demotion set made up of units of storage allocated to the storage group from the high performance disk group that have load scores that are less than or equal to the demotion threshold,

migrating host data stored on the units of storage in the demotion set to units of storage allocated from the lower performance disk group, and

deallocating, by the storage processor, the units of storage in the demotion set so that they are available for subsequent reallocation by the storage processor to store host data subsequently received by the storage processor.

2. The method of claim 1 , wherein detecting that the average response time for the storage group is exceeding the service level objective associated with the storage group includes calculating an average response time for the storage group during a preceding time period equal to a sum of the response times of the physical disk groups, wherein the response time of each physical disk group is weighted by the number of I/O operations performed per second during the preceding time period on storage units allocated to the storage group from that physical disk group.

3. The method of claim 2 , wherein detecting that the average response time for the storage group is exceeding the service level objective associated with the storage group further includes calculating a safety margin for the storage group that is equal to the product of multiplying i) the I/O operations performed per second on all storage units allocated to the storage group, by ii) a difference between the service level objective associated with the storage object and the average response time for the storage group during the preceding time period.

4. The method of claim 3 , wherein detecting that the average response time for the storage group is exceeding the service level objective associated with the storage group detects that the average response time for the storage group is exceeding the service level objective associated with the storage group only in response to the average response time for the storage group during a preceding time period being less than the service level objective associated with the storage group minus the safety margin for the storage group.

5. The method of claim 4 , wherein the high performance disk drives in the high performance disk group consist of at least one solid state drive; and

wherein the lower performance disk drives in the lower performance disk group consist of at least one magnetic hard disk drive.

6. The method of claim 5 , wherein migrating the host data stored on the units of storage in the demotion set to units of storage allocated from the lower performance disk group includes allocating units of storage from the lower performance disk group to store the host data stored on the units of storage in the demotion set.

7. The method of claim 6 , wherein the storage objects in the storage group comprise logical disks (LUNS) that are used by the application to store host data, and further comprising:

after deallocating the units of storage in the demotion set, allocating the units of storage in the demotion set to a storage object contained in another storage group in response to detecting that the other storage group is missing a service level objective associated with that storage group, wherein the storage object contained in the other storage group is a logical disk (LUN).

8. The method of claim 1 , wherein the storage objects in the storage group comprise virtual volumes that are used by the application to store host data.

9. The method of claim 1 , wherein the storage objects in the storage group comprise host file systems that are used by the application to store host data.

10. A computerized apparatus, comprising:

processing circuitry in a storage processor of a data storage system;

a memory coupled to the processing circuitry, the memory storing program code for proactively releasing high performance storage resources which, when executed by the processing circuitry, causes the processing circuitry to:

allocate units of storage from multiple physical disk groups to one or more storage objects in a storage group, wherein the storage objects are used by at least one application to store host data, wherein the multiple physical disk groups include a high performance disk group made up of one or more high performance disk drives and at least one lower performance disk group made up of one or more lower performance disk drives, wherein the storage group is associated with a service level objective, and wherein the service level objective comprises an average response time target for I/O operations performed on the storage objects in the storage group;

calculate a load score for each one of the units of storage allocated to the storage group from the high performance disk group;

determine a minimum load score equal to the lowest load score among the load scores calculated for the units of storage allocated from the high performance disk group to the storage group;

detect that an average response time for the storage group is exceeding the service level objective associated with the storage group; and

in response to detecting that the average response time for the storage group is exceeding the service level objective associated with the storage group,

calculate a demotion threshold based on the minimum load score, wherein the demotion threshold is a product of multiplying the minimum load score and a demotion range factor,

identify a demotion set made up of units of storage allocated to the storage group from the high performance disk group that have load scores that are less than or equal to the demotion threshold,

migrate host data stored on the units of storage in the demotion set to units of storage allocated from the lower performance disk group, and

deallocate the units of storage in the demotion set so that they are available for subsequent reallocation by the storage processor to store host data subsequently received by the storage processor.

11. The computerized apparatus of claim 10 , wherein the program code, when executed by the processing circuitry, further causes the processing circuitry to detect that the average response time for the storage group is exceeding the service level objective associated with the storage group in part by calculating an average response time for the storage group during a preceding time period equal to a sum of the response times of the physical disk groups, wherein the response time of each physical disk group is weighted by the number of I/O operations performed per second during the preceding time period on storage units allocated to the storage group from that physical disk group.

12. The computerized apparatus of claim 11 , wherein the program code, when executed by the processing circuitry, further causes the processing circuitry to detect that the average response time for the storage group is exceeding the service level objective associated with the storage group at least in part by calculating a safety margin for the storage group that is equal to the product of multiplying i) the I/O operations performed per second on all storage units allocated to the storage group, by ii) a difference between the service level objective associated with the storage object and the average response time for the storage group during the preceding time period.

13. The computerized apparatus of claim 12 , wherein the program code, when executed by the processing circuitry, further causes the processing circuitry to detect that the average response time for the storage group is exceeding the service level objective associated with the storage group by detecting that the average response time for the storage group is exceeding the service level objective associated with the storage group only in response to the average response time for the storage group during a preceding time period being less than the service level objective associated with the storage group minus the safety margin for the storage group.

14. The computerized apparatus of claim 13 , wherein the high performance disk drives in the high performance disk group consist of at least one solid state drive; and

wherein the lower performance disk drives in the lower performance disk group consist of at least one magnetic hard disk drive.

15. The computerized apparatus of claim 14 , wherein the program code, when executed by the processing circuitry, further causes the processing circuitry to migrate the host data stored on the units of storage in the demotion set to units of storage allocated from the lower performance disk group at least in part by allocating units of storage from the lower performance disk group to store the host data stored on the units of storage in the demotion set.

16. The computerized apparatus of claim 15 , wherein the storage objects in the storage group comprise logical disks (LUNS) that are used by the application to store host data; and

wherein the program code, when executed by the processing circuitry, further causes the processing circuitry to, after the units of storage in the demotion set are deallocated, allocate the units of storage in the demotion set to a storage object contained in another storage group in response to detecting that the other storage group is missing a service level objective associated with that storage group, wherein the storage object contained in the other storage group is a logical disk (LUN).

17. The computerized apparatus of claim 10 , wherein the storage objects in the storage group comprise virtual volumes that are used by the application to store host data.

18. The computerized apparatus of claim 10 , wherein the storage objects in the storage group comprise host file systems that are used by the application to store host data.

19. A computer program product, comprising:

a non-transitory computer readable medium storing program code for proactively releasing high performance storage resources, the set of instructions, when carried out by at least one processor in a storage processor in a data storage system, causing the processor to perform a method of:

allocating units of storage from multiple physical disk groups to one or more storage objects in a storage group, wherein the storage objects are used by at least one application to store host data, wherein the multiple physical disk groups include a high performance disk group made up of one or more high performance disk drives and at least one lower performance disk group made up of one or more lower performance disk drives, wherein the storage group is associated with a service level objective, and wherein the service level objective comprises an average response time target for I/O operations performed on the storage objects in the storage group;

calculating a load score for each one of the units of storage allocated to the storage group from the high performance disk group;

determining a minimum load score equal to the lowest load score among the load scores calculated for the units of storage allocated from the high performance disk group to the storage group;

detecting that an average response time for the storage group is exceeding the service level objective associated with the storage group; and

in response to detecting that the average response time for the storage group is exceeding the service level objective associated with the storage group,

calculating a demotion threshold based on the minimum load score, wherein the demotion threshold is a product of multiplying the minimum load score and a demotion range factor,

identifying a demotion set made up of units of storage allocated to the storage group from the high performance disk group that have load scores that are less than or equal to the demotion threshold,

migrating host data stored on the units of storage in the demotion set to units of storage allocated from the lower performance disk group, and

deallocating the units of storage in the demotion set so that they are available for subsequent reallocation by the storage processor to store host data subsequently received by the storage processor.