Flash cache partitioning
A computer implemented method, system, and computer program product for cache management comprising recording metadata of IO sent from the server to a storage array, calculating a distribution of a server cache based on the metadata, receiving an IO directed to the storage array, and revising an allocation of the server cache to the plurality of storage mediums based on the calculated distribution and the IO.
1. A system comprising:
a server having a cache;
a storage array, the storage array having a plurality of storage mediums and a storage cache and storing data; and
computer-executable logic operating in memory, wherein the computer-executable program logic is configured to enable one or more processors to execute:
recording metadata of Input\Output (IO) sent from the server to a storage array;
calculating a distribution of a server cache to the storage mediums based on the metadata;
receiving an IO directed to the storage array; and
revising an allocation of the server cache to the plurality of storage mediums based on the calculated distribution and the IO.
2. The system of claim 1 wherein the calculated distribution determines a cache allocation which minimizes the number of read misses in the server cache.
3. The system of claim 1 wherein the calculate distribution determines a cache allocation which maximizes the read hits in the server cache.
4. The system of claim 1 where the metadata recorded is large enough for the calculation of the distribution to determine whether the IO to the plurality of storage mediums is a random pattern or a predicable pattern.
5. The system of claim 4 wherein the server cache will not be assigned to random read patterns.
6. The system of claim 1 where the metadata recorded is large enough for the calculation of the distribution to determine what percentage of IO is directed to what portion of each of the plurality of storage mediums.
7. The system of claim 1 where the cache partitioning is recalculated based on the statistics at predetermined amounts of time.
8. A computer implemented method comprising:
recording metadata of Input\Output (IO) sent from a server to a storage array;
wherein the storage array comprises storage mediums;
calculating a distribution of a server cache to the storage mediums based on the metadata;
receiving an IO directed to the storage array; and
revising an allocation of the server cache to the plurality of storage mediums based on the calculated distribution and the IO.
9. The method of claim 8 wherein the calculated distribution determines a cache allocation which minimizes the number of read misses in the server cache.
10. The method of claim 8 wherein the calculate distribution determines a cache allocation which maximizes the read hits in the server cache.
11. The method of claim 8 where the metadata recorded is large enough for the calculation of the distribution to determine whether the IO to the plurality of storage mediums is a random pattern or a predicable pattern.
12. The method of claim 11 wherein the server cache will not be assigned to random read patterns.
13. The method of claim 8 where the metadata recorded is large enough for the calculation of the distribution to determine what percentage of IO is directed to what portion of each of the plurality of storage mediums.
14. A computer program product comprising:
a non-transitory computer readable medium encoded with computer executable program, the code enabling:
recording metadata of Input\Output (IO) sent from a server to a storage array; wherein the storage array comprises storage mediums;
calculating a distribution of a server cache to the storage mediums based on the metadata;
receiving an IO directed to the storage array; and
revising an allocation of the server cache to the plurality of storage mediums based on the calculated distribution and the IO.
15. The computer program product of claim 14 wherein the calculated distribution determines a cache allocation which minimizes the number of read misses in the server cache.
16. The computer program product of claim 14 wherein the calculate distribution determines a cache allocation which maximizes the read hits in the server cache.
17. The computer program product of claim 14 where the metadata recorded is large enough for the calculation of the distribution to determine whether the IO to the plurality of storage mediums is a random pattern or a predicable pattern.
18. The computer program product of claim 17 wherein the server cache will not be assigned to random read patterns.
19. The computer program product of claim 14 where the metadata recorded is large enough for the calculation of the distribution to determine what percentage of IO is directed to what portion of each of the plurality of storage mediums.
20. The computer program product of claim 14 where the cache partitioning is recalculated based on the statistics at predetermined amounts of time.