Cache utility modeling for automated cache configuration

1. An apparatus comprising:

circuitry; and

logic for execution by the circuitry to:

monitor processing of Input/Output (I/O) requests of an application being executed by a computing platform by collecting a trace of the I/O requests, the trace including an I/O class of each I/O request,

replay the trace and automatically analyze possible cache configuration policies for using a cache during execution of the application by the computing platform,

determine a cache utility metric for each I/O class;

rank I/O classes according to the cache utility metrics from most useful to least useful for improving performance of the computing platform while executing the application; and

determine an optimal cache configuration policy for the cache from the possible cache configuration policies.

2. The apparatus of claim 1 , comprising the logic to:

apply the optimal cache configuration policy to use of the cache during subsequent execution of the application by the computing platform.

3. The apparatus of claim 1 , comprising the logic to:

determine the optimal cache configuration policy based on the I/O class rankings.

4. The apparatus of claim 3 , comprising the logic to:

apply a “cache all” policy and an infinite cache size when replaying the trace and automatically analyze possible cache configuration policies for using the cache.

5. The apparatus of claim 1 , wherein the cache utility metric comprises a number of disk seeks for a first I/O class divided by a number of sectors accessed by the I/O first class.

6. The apparatus of claim 1 , wherein the cache utility metric comprises a number of disk seeks for a second I/O class divided by a number of I/O requests for the I/O second class.

7. The apparatus of claim 1 , comprising the logic to:

perform the monitoring, replaying, and determining when a workload of the application changes.

8. A method comprising:

monitoring processing of I/O requests of an application being executed by a computing platform by collecting a trace of the I/O requests, the trace including an I/O class of each I/O request;

replaying the trace and automatically analyzing possible cache configuration policies for using a cache during execution of the application by the computing platform;

determining a cache utility metric for each I/O class;

ranking I/O classes according to the cache utility metrics from most useful to least useful for improving performance of the computing platform while executing the application; and

determining an optimal cache configuration policy for the cache from the possible cache configuration policies.

9. The method of claim 8 , comprising:

applying the optimal cache configuration policy to use of the cache during subsequent execution of the application by the computing platform.

10. The method of claim 8 , comprising:

determining the optimal cache configuration policy based on the I/O class rankings.

11. The method of claim 10 , comprising:

applying a “cache all” policy and an infinite cache size when replaying the trace and automatically analyze possible cache configuration policies for using the cache.

12. The method of claim 8 , wherein the cache utility metric comprises a number of disk seeks for a first I/O class divided by a number of sectors accessed by the first I/O class.

13. The method of claim 8 , wherein the cache utility metric comprises a number of disk seeks for a second I/O class divided by a number of I/O requests for the I/O second class.

14. The method of claim 8 , comprising:

performing the monitoring, replaying, and determining when a workload of the application changes.

15. At least one non-transitory machine readable medium comprising a plurality of instructions that in response to being executed by a system at a computing platform cause the system to:

monitor processing of I/O requests of an application being executed by a computing platform by collecting a trace of the I/O requests, the trace including an I/O class of each I/O request,

replay the trace and automatically analyze possible cache configuration policies for using a cache during execution of the application by the computing platform,

determine a cache utility metric for each I/O class;

rank I/O classes according to the cache utility metrics from most useful to least useful for improving performance of the computing platform while executing the application; and

determine an optimal cache configuration policy for the cache from the possible cache configuration policies.

16. The at least one non-transitory machine readable medium of claim 15 , comprising the instructions to further cause the system to:

apply the optimal cache configuration policy to use of the cache during subsequent execution of the application by the computing platform.

17. The at least one non-transitory machine readable medium of claim 15 , comprising the insertions to further cause the system to:

determine the optimal cache configuration policy based on the I/O class rankings.

18. The at least one non-transitory machine readable medium of claim 17 , comprising the insertions to further cause the system to:

apply a “cache all” policy and an infinite cache size when replaying the trace and automatically analyze possible cache configuration policies for using the cache.

19. The at least one non-transitory machine readable medium of claim 15 , wherein the cache utility metric comprises a number of disk seeks for a first I/O class divided by a number of sectors accessed by the first I/O class.

20. The at least one non-transitory machine readable medium of claim 15 , wherein the cache utility metric comprises a number of disk seeks for a second I/O class divided by a number of I/O requests for the second I/O class.

21. The at least one non-transitory machine readable medium of claim 15 , comprising the insertions to further cause the system to:

perform the monitoring, replaying, and determining when a workload of the application changes.