Cache memory management
One or more aspects of the present disclosure relate to cache memory management. In embodiments, a global memory of a storage array into one or more cache partitions based on an anticipated activity of one or more input/output (IO) service level (SL) workload volumes can be dynamically partitioned.
1. An apparatus comprising a memory and at least one processor configured to:
dynamically partition a global memory of a storage array into one or more cache partitions based on an anticipated activity of one or more input/output (IO) service level (SL) workload volumes, wherein dynamically partitioning the global memory includes partitioning the global memory along a first dimension corresponding to IO type and a second dimension corresponding to track sizes of IO operations in an IO workload;
dynamically establish SL cache groups from the partitioned global memory based on an anticipated activity, wherein SL cache group establishment includes dynamically setting a bit in each global memory cache slot, defining each cache slot's assigned SL cache group; and
increase an amount of global memory cache slots assigned to process critical IOs or limit non-critical IO consumption of the global memory cache slots based on a quality of service (QoS) corresponding to each IO service level.
2. The apparatus of claim 1 further configured to:
receive an IO workload; and
analyze one or more SL workload volumes received within the IO workload.
3. The apparatus of claim 2 , wherein the IO workload includes one or more IO operations, wherein each IO operation includes at least one of: a read request and a write request.
4. The apparatus of claim 1 further configured to:
determine a current state of the global memory; and
identify historical IO patterns of the IO workloads previously received by the storage array.
5. The apparatus of claim 1 further configured to:
determine a current state of the global memory; and
identify an SL workload usage pattern of the global memory based on an IO SL workload usage history of the global memory.
6. The apparatus of claim 5 further configured to:
determine a current state of each cache slot of the global memory; and
identify an IO operation SL tier usage pattern of each cache slot based on a history of IO SL tier usage of each cache slot, wherein each SL tier corresponds to an expected performance metric related to processing an IO operation of each SL tier by the storage array.
7. The apparatus of claim 6 further configured to generate an SL trend model corresponding to the SL tier usage pattern using a machine learning engine configured to process IO workloads using one or more pattern learning techniques such as an inference learning technique.
8. The apparatus of claim 7 further configured to dynamically partition the global memory into one or more cache banks based on the SL trend model.
9. The apparatus of claim 8 further configured to use the SL trend model to control usage of each cache slot and/or cache bank by incoming IO operations.
10. The apparatus of claim 8 further configured to:
dynamically assign each cache slot to cache IO operations of a particular SL tier by modifying metadata of each cache; and
establish a time threshold during which each cache slot can be exclusively assigned to cache IO operations associated with any particular SL tier.
11. A method comprising:
dynamically partition a global memory of a storage array into one or more cache partitions based on an anticipated activity of one or more input/output (IO) service level (SL) workload volumes, wherein dynamically partitioning the global memory includes partitioning the global memory along a first dimension corresponding to IO type and a second dimension corresponding to track sizes of IO operations in an IO workload;
dynamically establishing SL cache groups from the partitioned global memory based on an anticipated activity, wherein SL cache group establishment includes dynamically setting a bit in each global memory cache slot, defining each cache slot's assigned SL cache group; and
increasing an amount of global memory cache slots assigned to process critical IOs or limiting non-critical IO consumption of the global memory cache slots based on a quality of service (QoS) corresponding to each IO service level.
12. The method of claim 11 further comprising:
receiving an IO workload; and
analyzing one or more SL workload volumes received within the IO workload.
13. The method of claim 12 , wherein the IO workload includes one or more IO operations, wherein each IO operation includes at least one of: a read request and a write request.
14. The method of claim 11 further comprising:
determining a current state of the global memory; and
identifying historical IO patterns of the IO workloads previously received by the storage array.
15. The method of claim 11 further comprising:
determining a current state of the global memory; and
identifying an SL workload usage pattern of the global memory based on an IO SL workload usage history of the global memory.
16. The method of claim 15 further comprising:
determining a current state of each cache slot of the global memory; and
identifying an IO operation SL tier usage pattern of each cache slot based on a history of IO SL tier usage of each cache slot, wherein each SL tier corresponds to an expected performance metric related to processing an IO operation of each SL tier by the storage array.
17. The method of claim 16 further comprising generating an SL trend model corresponding to the SL tier usage pattern using a machine learning engine including process IO workloads using one or more pattern learning techniques such as an inference learning technique.
18. The method of claim 17 further comprising dynamically partitioning the global memory into one or more cache banks based on the SL trend model.
19. The method of claim 18 further comprising using the SL trend model to control usage of each cache slot and/or cache bank by incoming IO operations.
20. The method of claim 18 further comprising:
dynamically assigning each cache slot to cache IO operations of a particular SL tier by modifying metadata of each cache; and
establishing a time threshold during which each cache slot can be exclusively assigned to cache IO operations associated with any particular SL tier.