Persistent memory adaptation
Embodiments of the present disclosure relate to establishing persistent cache memory as a write tier. An input/output (IO) workload of a storage array can be analyzed. One or more write data portions of the IO workload can be stored in a persistent memory region of one or more disks of the storage array.
1. An apparatus comprising a memory and at least one processor configured to:
analyze an input/output (IO) workload of a storage array;
in a backend portion of the storage array:
establish at least one logical drive corresponding to one or more address spaces within a persistent data storage device of the storage array, wherein the address spaces correspond to a physical location within the persistent data storage device;
identify each address space with a write request density above a threshold;
allocate a persistent memory region (PMR) of persistent cache memory within the persistent data storage device to cache write data requests targeting the address spaces with the write request densities above the threshold;
establish a virtualized memory storage corresponding to a persistent memory region (PMR) of persistent cache memory within the persistent data storage device; and
direct IO write requests to the at least one logical drive or the virtualized memory storage based on the identified address spaces with write request densities above the threshold.
2. The apparatus of claim 1 further configured to provision the persistent memory region of the one or more disks to store the one or more write data portions.
3. The apparatus of claim 2 , wherein the persistent memory region includes non-volatile storage.
4. The apparatus of claim 2 , wherein the persistent memory region is configured with a similar RAID striping as block storage devices of the storage array.
5. The apparatus of claim 1 further configured to identify write operations included in the IO workloads.
6. The apparatus of claim 5 , wherein identifying the write operations includes configuring an IO operation filter to identify write operations within the IO workload based on metadata associated with each IO operation in the IO workload.
7. The apparatus of claim 5 further configured to identify write densities of address spaces corresponding to the write operations.
8. The apparatus of claim 7 further configured to perform one or more machine learning techniques to anticipate write densities of the address spaces, wherein the one or more machine learning techniques includes an instance-based learning processor.
9. The apparatus of claim 8 further configured to:
generate a data structure identifying address spaces organized by an activity ranking metric that identifies write activity levels of each address space; and
generate one or more traffic policies configured to store those IO write operations and their corresponding data having a ranking metric above a threshold rank to the persistent memory region.
10. The apparatus of claim 9 further configured to direct IO write operations to the persistent memory region using the one or more traffic policies.
11. A method comprising:
analyzing an input/output (IO) workload of a storage array;
in a backend portion of the storage array:
establishing at least one logical drive corresponding to one or more address spaces within a persistent data storage device of the storage array, wherein the address spaces correspond to a physical location within the persistent data storage device;
identifying each address space with a write request density above a threshold;
allocating a persistent memory region (PMR) of persistent cache memory within the persistent data storage device to cache write data requests targeting the address spaces with the write request densities above the threshold;
establishing a virtualized memory storage corresponding to a persistent memory region (PMR) of persistent cache memory within the persistent data storage device; and
directing IO write requests to the at least one logical drive or the virtualized memory storage based on the identified address spaces with write request densities above the threshold.
12. The method of claim 1 further comprising provisioning the persistent memory region of the one or more disks to store the one or more write data portions.
13. The method of claim 2 , wherein the persistent memory region includes non-volatile storage.
14. The method of claim 2 , wherein the persistent memory region is configured with a similar RAID striping as block storage devices of the storage array.
15. The method of claim 1 further comprising identifying write operations included in the IO workloads.
16. The method of claim 5 , wherein identifying the write operations includes configuring an IO operation filter to identify write operations within the IO workload based on metadata associated with each IO operation in the IO workload.
17. The method of claim 5 further comprising identifying write densities of address spaces corresponding to the write operations.
18. The method of claim 7 further comprising performing one or more machine learning techniques to anticipate write densities of the address spaces, wherein the one or more machine learning techniques includes an instance-based learning processor.
19. The method of claim 8 further comprising:
generating a data structure identifying address spaces organized by an activity ranking metric that identifies write activity levels of each address space; and
generating one or more traffic policies comprising store those IO write operations and their corresponding data having a ranking metric above a threshold rank to the persistent memory region.
20. The method of claim 9 further comprising directing IO write operations to the persistent memory region using the one or more traffic policies.