Dynamic management of garbage collection and overprovisioning for host stream storage

• US 10,452,536 B2
• Filed: 10/13/2017
• Issued: 10/22/2019
• Est. Priority Date: 06/12/2017
• Status: Active Grant

• Current Assignee: JP Morgan Chase Bank, N.A. (JP Morgan Chase & Co)
• Original Assignee: Western Digital Technologies Incorporated (Western Digital Corporation)
• Inventors: Parker, Liam Michael, Gorobets, Sergey Anatolievich, Acosta, Marc
• Primary Examiner(s): Bataille, Pierre Miche
• Application Number:
  US15/783,739
• Publication Number:
  US 20180357159A1
• Time in Patent Office: 739 Days
• Field of Search: 711165, 711162, 711203, 711170
• CPC Class Codes: G06F 12/0246 in block erasable memory, e.g. flash memory
• CPC Class Codes: G06F 12/0253 Garbage collection, i.e. reclamation of unreferenced memory
• CPC Class Codes: G06F 2212/702 Conservative garbage collection
• CPC Class Codes: G06F 2212/7205 Cleaning, compaction, garbage collection, erase control
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0616 in relation to life time, e.g. increasing Mean Time Between Failures [MTBF]
• CPC Class Codes: G06F 3/0631 by allocating resources to storage systems
• CPC Class Codes: G06F 3/0653 Monitoring storage devices or systems
• CPC Class Codes: G06F 3/0655 Vertical data movement, i.e. input-output transfer; data movement between one or more hosts and one or more storage devices
• CPC Class Codes: G06F 3/0688 Non-volatile semiconductor memory arrays

Abstract

A system and method is disclosed for managing data in a non-volatile memory. The system may include a non-volatile memory having multiple non-volatile memory sub-drives. A controller of the memory system is configured receive data streams from multiple different host systems and keep data for the separate streams in separate sub-drives. The method may include dynamically changing overprovisioning of the sub-drives in response to changes in relative workload measurements of data writes coming from the different host systems.

Claims

1. A method for managing data in a non-volatile memory system having a controller in communication with a plurality of sub-drives, the method comprising the controller:

receiving host data writes of a plurality of data streams from a plurality of host sources at the memory system, wherein each of the plurality of data streams is associated with a respective tenant, each respective tenant corresponding to a host system of a different remotely located subscriber, wherein each tenant is associated with a minimum amount of overprovisioning;

for each of the plurality of data streams associated with the respective tenant, storing all received host writes associated with the respective tenant only in a respective one sub-drive of the plurality of sub-drives, the respective one sub-drive being only utilized for the respective tenant, wherein each of the plurality of data streams is separately identifiable based at least in part on identification information that is included with each stream write from the respective tenant;

measuring a host write workload of each of the plurality of sub-drives; and

altering an amount of overprovisioning of one of the plurality of sub-drives based on the minimum amount of overprovisioning associated with the respective tenant of the sub-drive and a change in the measured host write workload of another of the plurality of sub-drives.

2. The method of claim 1, wherein altering the amount of overprovisioning of the one of the plurality of sub-drives comprises:

selecting a source superblock from the one of the plurality of sub-drives;

relocating valid data from the selected source superblock only into a relocation superblock in the one of the plurality of sub-drives; and

assigning the source superblock to a shared free block pool for the non-volatile memory system after relocating the valid data.

3. The method of claim 1, wherein measuring the host write workload comprises:

detecting a garbage collection trigger; and

in response to detecting the garbage collection trigger determining a percentage of host writes sent to each of the plurality of sub-drives for a predetermined window of time.

4. The method of claim 3, wherein altering the amount of overprovisioning of the one of the plurality of sub-drives based on the minimum amount of overprovisioning associated with the respective tenant of the sub-drive and the change in the measured host write workload of another of the plurality of sub-drives comprises:

for each respective sub-drive, determining a target overprovisioning amount based on the determined percentage of host writes sent to the respective sub-drive, the minimum amount of overprovisioning associated with the respective tenant, and a percentage of a total valid logical space of the memory system in the respective sub-drive; and

performing a garbage collection operation in a selected sub-drive of the plurality of sub-drives, wherein the selected sub-drive comprises a sub-drive having a greatest amount of actual overprovisioning above the target overprovisioning amount determined for that sub-drive.

5. The method of claim 4, wherein performing the garbage collection operation comprises:

relocating valid data from a selected source superblock in the selected sub-drive only into a relocation superblock in the selected sub-drive; and

assigning the source superblock to a shared free block pool for the non-volatile memory system after relocating the valid data.

6. The method of claim 5, further comprising selecting a superblock having a least amount of valid data in the selected sub-drive as the selected source superblock.

7. The method of claim 3, wherein altering the amount of overprovisioning of the one of the plurality of sub-drives based on the minimum amount of overprovisioning associated with the respective tenant of the sub-drive and the change in the measured host write workload of another of the plurality of sub-drives comprises:

for each respective sub-drive of the plurality of sub-drives determining a target overprovisioning amount based on the determined percentage of host writes sent to the respective sub-drive, and based on a percentage of a total valid logical space of the memory system in the respective sub-drive; and

performing a garbage collection operation in a selected sub-drive of the plurality of sub-drives, wherein the selected sub-drive comprises a sub-drive having a greatest amount of actual overprovisioning above a greater of the target overprovisioning amount determined for that sub-drive or any predetermined minimum overprovisioning associated with the sub-drive.

8. The method of claim 1, further comprising the controller storing a recent current workload calculation for the one for the plurality of sub-drives when the one of the plurality of sub-drives becomes idle.

9. The method of claim 8, further comprises the controller:

retrieving the recent current workload calculation for the one for the plurality of sub-drives, when the one of the plurality of sub-drives becomes active after being idle; and

applying the recent current workload calculation as a current workload calculation to a new overprovisioning determination for all of the plurality of sub-drives.

10. A non-volatile memory system for managing data from a plurality of hosts, the system comprising:

a non-volatile memory having a plurality of sub-drives; and

a controller in communication with the plurality of sub-drives, the controller configured to;

receive host data writes of a plurality of data streams from the plurality of hosts, wherein each of the plurality of data streams is associated with a respective tenant, each respective tenant corresponding to a host system of a different remotely located subscriber, wherein each tenant is associated with a minimum amount of overprovisioning;

maintain data from each of the plurality of data streams in a different respective separate sub-drive, the different respective separate sub-drive being only utilizable for the respective tenant, each of the plurality of data streams being separately identifiable based at least in part on identification information that is included with each stream write from the respective tenant;

measure a host write workload of each of the plurality of sub-drives;

determine a respective target overprovisioning level for each of the plurality of sub-drives based on a relative host write workload of each of the plurality of sub-drives; and

adjust an overprovisioning of one of the plurality of sub-drives in accordance with the respective target overprovisioning level determined for the one of the plurality of sub-drives and the minimum amount of overprovisioning associated with the respective tenant of the sub-drive.

11. The non-volatile memory system of claim 10, wherein the non-volatile memory comprises a substrate formed with a three-dimensional memory structure.

12. The non-volatile memory system of claim 10, wherein to adjust the overprovisioning, the controller is configured to perform a garbage collection operation on the one of the plurality of sub-drives.

13. The non-volatile memory system of claim 12, wherein to perform a garbage collection operation to adjust the overprovisioning, the controller is configured to:

select a source superblock from the one of the plurality of sub-drives;

relocate valid data from the selected source superblock only into a relocation superblock in the one of the plurality of sub-drives; and

after relocating the valid data from the selected source superblock, assign the source superblock to a shared free block pool for the non-volatile memory system.

14. The non-volatile memory system of claim 10, wherein to measure the host write workload the controller is further configured to:

detect a garbage collection trigger; and

in response to detecting the garbage collection trigger, determine a current workload of an amount of host writes for a predetermined window of time for each of the plurality of sub-drives.

15. The non-volatile memory system of claim 10, wherein the controller is configured to, in response to one of the plurality of sub-drives becoming idle, store a recent current workload calculation for the one for the plurality of sub-drives.

16. The non-volatile memory system of claim 15, wherein the controller is configured to:

retrieve the recent current workload calculation stored for the one for the plurality of sub-drives when the one of the plurality of sub-drives becomes active after being idle; and

apply the recent current workload calculation as a current workload calculation to a new overprovisioning determination for all of the plurality of sub-drives.

17. The non-volatile memory system of claim 10, wherein the controller is configured to determine the respective target overprovisioning level for each of the plurality of sub-drives based on a percentage of host writes sent to each respective sub-drive, and based on a percentage of a total valid logical space of the non-volatile memory system assigned to the respective sub-drive;

and

wherein to adjust the overprovisioning of one of the plurality of sub-drives, the controller is configured to;

select a sub-drive having a greatest amount of actual overprovisioning above the respective target overprovisioning level determined for that sub-drive; and

perform a garbage collection operation on the selected sub-drive.

18. The non-volatile memory system of claim 17, wherein the controller is configured to determine the respective target overprovisioning level for each of the plurality of sub-drives according to a relation:

19. A non-volatile memory system for managing data from a plurality of hosts, the system comprising:

a non-volatile memory having a plurality of sub-drives;

means for receiving host data writes of a plurality of different data streams from a plurality of host sources at the memory system and for storing all received host writes for each particular different data stream only in a respective one of the plurality of sub-drives, wherein each of the plurality of different data streams is associated with a respective tenant, each respective tenant corresponding to a host system of a different remotely located subscriber, the respective one of the plurality of sub-drives being only utilizable for the respective tenant, each of the plurality of different data streams being separately identifiable based at least in part on identification information that is included with each stream write from the respective tenant, wherein each tenant is associated with a minimum amount of overprovisioning;

means for measuring a host write workload of each of the plurality of sub-drives; and

means for altering an amount of overprovisioning of one of the plurality of sub-drives based on the minimum amount of overprovisioning associated with the respective tenant of the sub-drive and a change in the measured host write workload of another of the plurality of sub-drives.

20. The non-volatile memory system of claim 19, wherein the means for altering the amount of overprovisioning comprises:

means for determining a target overprovisioning amount in a particular sub-drive based on a determined percentage of host writes sent to the particular sub-drive, the minimum amount of overprovisioning associated with the respective tenant, and on a percentage of a total valid logical space of the memory system in the particular sub-drive; and

means for performing a garbage collection operation in a selected sub-drive of the plurality of sub-drives, wherein the selected sub-drive comprises a sub-drive having a greatest amount of actual overprovisioning above the target overprovisioning amount determined for that sub-drive.