Dynamic enabling of metadata prefetch into cache

1. A method for use in a storage system, comprising:

defining a first plurality of time windows, the first plurality of time windows including a current time window and one or more past time windows;

calculating a first plurality of weights, each of the first plurality of weights corresponding to a different one of the first plurality of time windows;

identifying a first plurality of values of a cache metric, wherein each of the first plurality of values of the cache metric corresponds to a different one of the first plurality of time windows;

calculating a prefetch score for a first type of data based on the first plurality of weights and the first plurality of caching metric values, the prefetch score being calculated by weighing each of the cache metric values based on a respective one of the first plurality of weights that corresponds to a same time window as the cache metric value; and

when the prefetch score fails to meet a threshold, stopping prefetching of the first type of data, while continuing to prefetch a second type of data.

2. The method of claim 1 , wherein the prefetching of the first type of data is stopped by modifying a data structure that is stored in a memory of the storage system, the data structure being modified to indicate that the prefetching of the first type of data is disabled.

3. The method of claim 1 , wherein the first type of data includes a first type of metadata that is used by the storage system for storing user data, and the second type of data includes a second type of metadata that is used by the storage system for storing user data.

4. The method of claim 1 , wherein the first type of data includes user data that is stored in a first logic unit (LU) of the storage system and the second type of data includes user data that is stored in a second LU of the storage system.

5. The method of claim 1 , wherein the first plurality of weights is generated by using an online machine learning engine that is configured to balance: (i) reductions in prefetching that is performed by the storage system with (ii) an increase in cache misses that occur in the storage system as a result of the reductions in prefetching, the online machine learning engine being executed by the storage system concurrently with a servicing of incoming Input/Output (I/O) requests by the storage system.

6. The method of claim 1 , further comprising recalculating the prefetch score for the first type of data and resuming the prefetching of the first type of data when the recalculated prefetch score for the first type of data meets the threshold, the prefetch score being recalculated at runtime of the storage system.

7. The method of claim 6 , further comprising:

defining a second plurality of time windows;

calculating a second plurality of weights, each of the second plurality of weights corresponding to a different one of the second plurality of time windows;

identifying a second plurality of values of a cache metric, wherein each of the second plurality of values of the cache metric corresponds to a different one of the second plurality of time windows,

wherein the prefetch score is recalculated based on the second plurality of weights and the second plurality of caching metric values.

8. A storage system comprising:

a memory; and

one or more processors operatively coupled to the memory, the one or more processors being configured to perform the operations of:

defining a first plurality of time windows, the first plurality of time windows including a current time window and one or more past time windows;

calculating a first plurality of weights, each of the first plurality of weights corresponding to a different one of the first plurality of time windows;

identifying a first plurality of values of a cache metric, wherein each of the first plurality of values of the cache metric corresponds to a different one of the first plurality of time windows;

calculating a prefetch score for a first type of data based on the first plurality of weights and the first plurality of caching metric values, the prefetch score being calculated by weighing each of the cache metric values based on a respective one of the first plurality of weights that corresponds to a same time window as the cache metric value; and

when the prefetch score fails to meet a threshold, stopping prefetching of the first type of data, while continuing to prefetch a second type of data.

9. The storage system of claim 8 , wherein:

the memory is configured to store a data structure including a plurality of entries, each entry including a respective identifier corresponding to a different respective type of data that is stored in the storage system and a respective indication of whether prefetching of the respective type of data is enabled,

the prefetching of the first type of data is stopped by modifying the data structure to indicate that the prefetching of the first type of data is disabled.

10. The storage system of claim 8 , wherein the first type of data includes a first type of metadata that is used by the storage system for storing user data, and the second type of data includes a second type of metadata that is used by the storage system for storing user data.

11. The storage system of claim 8 , wherein the first type of data includes user data that is stored in a first logic unit (LU) of the storage system and the second type of data includes user data that is stored in a second LU of the storage system.

12. The storage system of claim 8 , wherein the first plurality of weights is generated by using an online machine learning engine that is configured to balance: (i) reductions in prefetching that is performed by the storage system with (ii) an increase in cache misses that occur in the storage system as a result of the reductions in prefetching, the online machine learning engine being executed by the storage system concurrently with a servicing of incoming Input/Output (I/O) requests by the storage system.

13. The storage system of claim 8 , wherein the one or more processors are further configured to perform the operations of:

recalculating the prefetch score for the first type of data; and

resuming the prefetching of the first type of data when the recalculated prefetch score for the first type of data meets the threshold, the prefetch score being recalculated at runtime of the storage system.

14. The storage system of claim 13 , wherein:

the one or more processors are further configured to perform the operations of: defining a second plurality of time windows, calculating a second plurality of weights, each of the second plurality of weights corresponding to a different one of the second plurality of time windows, identifying a second plurality of values of a cache metric; and

the prefetch score is recalculated based on the second plurality of weights and the second plurality of caching metric values.

15. A non-transitory computer-readable medium storing one or more processor-executable instructions, which when executed by one or more processors, cause the one or more processors to perform the operations of:

defining a first plurality of time windows, the first plurality of time windows including a current time window and one or more past time windows;

calculating a first plurality of weights, each of the first plurality of weights corresponding to a different one of the first plurality of time windows;

identifying a first plurality of values of a cache metric, wherein each of the first plurality of values of the cache metric corresponds to a different one of the first plurality of time windows;

calculating a prefetch score for a first type of data based on the first plurality of weights and the first plurality of caching metric values, the prefetch score being calculated by weighing each of the cache metric values based on a respective one of the first plurality of weights that corresponds to a same time window as the cache metric value; and

when the prefetch score fails to meet a threshold, stopping prefetching of the first type of data, while continuing to prefetch a second type of data.

16. The non-transitory computer-readable medium of claim 15 , wherein the prefetching of the first type of data is stopped by modifying a data structure to indicate that the prefetching of the first type of data is disabled.

17. The non-transitory computer-readable medium of claim 15 , wherein the first type of data includes a first type of metadata that is used by a storage system for storing user data, and the second type of data includes a second type of metadata that is used by the storage system for storing user data.

18. The non-transitory computer-readable medium of claim 15 , wherein the first type of data includes user data that is stored in a first logic unit (LU) of a storage system and the second type of data includes user data that is stored in a second LU of the storage system.

19. The non-transitory computer-readable medium of claim 15 , wherein the first plurality of weights is generated by using an online machine learning engine that is configured to balance: (i) reductions in prefetching that is performed by a storage system with (ii) an increase in cache misses that occur in the storage system as a result of the reductions in prefetching.

20. The non-transitory computer-readable medium of claim 15 , wherein the processor-executable instructions, when executed by the one or more processors, further cause the one or more processors to prefetch score for the first type of data and resuming the prefetching of the first type of data when the recalculated prefetch score for the first type of data meets the threshold, the prefetch score being recalculated at runtime of a storage system.