Performance monitoring for storage system with core thread comprising internal and external schedulers

1. An apparatus comprising:

at least one processing device comprising a processor coupled to a memory;

said at least processing device being configured:

to monitor performance of a first thread of a first application executing on one of a plurality of processing cores of a storage system;

the first thread comprising an internal scheduler configured to control switching between a plurality of sub-threads of the first thread, and an external scheduler configured to control release of the processing core by the first thread for use by at least a second thread of a second application different than the first application;

in conjunction with monitoring the performance of the first thread in executing the first application, to maintain a cumulative suspend time of the first thread over multiple suspensions of the first thread, one or more of the multiple suspensions allowing at least the second thread of the second application to execute on the processing core; and

to generate performance measurements for respective ones of the sub-threads of the first thread based at least in part on the cumulative suspend time of the first thread.

2. The apparatus of claim 1 wherein said at least one processing device comprises at least a subset of the plurality of processing cores of the storage system.

3. The apparatus of claim 1 wherein the first application comprises a block-storage application of the storage system.

4. The apparatus of claim 1 wherein the second application comprises a file-storage application of the storage system.

5. The apparatus of claim 1 wherein the first thread comprises a truck thread that when executing on the processing core is configured to utilize substantially all available processing resources of the processing core.

6. The apparatus of claim 1 wherein the external scheduler of the first thread comprises:

a processing core release component configured to determine, in conjunction with each switch between sub-threads of the first thread, whether or not the first thread will suspend itself so as to release the processing core for use by at least the second thread of the second application; and

a waker component configured to determine, in conjunction with each switch between sub-threads of the first thread, whether or not there is at least one additional thread of the first application to be returned from suspension prior to release of the processing core by the first thread;

the processing core release component further operating in conjunction with the waker component to return the first thread from suspension.

7. The apparatus of claim 1 wherein each of the sub-threads of the first thread is in one of multiple designated states at a particular point in time, including a running state, a ready state and a suspended state.

8. The apparatus of claim 7 wherein the sub-threads comprise respective non-preemptive threads and the first thread is configured such that no sub-thread in the running state is suspended to allow release of the processing core by the first thread for use by the second thread, the multiple suspensions of the first thread each occurring in conjunction with a switch between sub-threads of the first thread.

9. The apparatus of claim 7 wherein generating performance measurements for respective ones of the sub-threads of the first thread based at least in part on the cumulative suspend time of the first thread comprises:

generating a ready time measurement for a particular one of the sub-threads; and

adjusting the ready time measurement based at least in part on the cumulative suspend time of the first thread;

wherein the adjusted ready time measurement is indicative of an amount of time the sub-thread spent in the ready state while the first thread was not suspended.

10. The apparatus of claim 7 wherein generating performance measurements for respective ones of the sub-threads of the first thread based at least in part on the cumulative suspend time of the first thread comprises:

generating a suspend time measurement for a particular one of the sub-threads; and

adjusting the suspend time measurement based at least in part on the cumulative suspend time of the first thread;

wherein the adjusted suspend time measurement is indicative of an amount of time the sub-thread spent in the suspended state while the first thread was not suspended.

11. The apparatus of claim 1 wherein maintaining a cumulative suspend time of the first thread over multiple suspensions of the first thread comprises, for each of the multiple suspensions of the first thread:

taking a first timestamp to indicate start of the suspension;

taking a second timestamp to indicate resumption of the first thread from the suspension; and

adding a time interval corresponding to a difference between the first and second timestamps to the cumulative suspend time.

12. The apparatus of claim 1 wherein maintaining a cumulative suspend time of the first thread over multiple suspensions of the first thread comprises, for each of the multiple suspensions of the first thread, storing a current value of the cumulative suspend time in association with a current value of a suspension number counter that is incremented with each of the multiple suspensions of the first thread.

13. The apparatus of claim 12 wherein generating performance measurements for respective ones of the sub-threads of the first thread based at least in part on the cumulative suspend time of the first thread comprises:

taking a first timestamp to indicate start of a performance interval;

storing a starting value of the suspension number counter in conjunction with taking the first timestamp;

taking a second timestamp to indicate end of the performance interval;

storing an ending value of the suspension number counter in conjunction with taking the second timestamp;

utilizing the starting and ending values of the suspension number counter to determine respective corresponding values of the cumulative suspend time of the first thread; and

adjusting a difference between the first and second timestamps based at least in part on a difference between the corresponding values of the cumulative suspend time of the first thread.

14. A method comprising:

monitoring performance of a first thread of a first application executing on one of a plurality of processing cores of a storage system;

the first thread comprising an internal scheduler configured to control switching between a plurality of sub-threads of the first thread, and an external scheduler configured to control release of the processing core by the first thread for use by at least a second thread of a second application different than the first application;

in conjunction with monitoring the performance of the first thread in executing the first application, maintaining a cumulative suspend time of the first thread over multiple suspensions of the first thread, one or more of the multiple suspensions allowing at least the second thread of the second application to execute on the processing core; and

generating performance measurements for respective ones of the sub-threads of the first thread based at least in part on the cumulative suspend time of the first thread.

15. The method of claim 14 wherein maintaining a cumulative suspend time of the first thread over multiple suspensions of the first thread comprises, for each of the multiple suspensions of the first thread:

taking a first timestamp to indicate start of the suspension;

taking a second timestamp to indicate resumption of the first thread from the suspension; and

adding a time interval corresponding to a difference between the first and second timestamps to the cumulative suspend time.

16. The method of claim 14 wherein maintaining a cumulative suspend time of the first thread over multiple suspensions of the first thread comprises, for each of the multiple suspensions of the first thread, storing a current value of the cumulative suspend time in association with a current value of a suspension number counter that is incremented with each of the multiple suspensions of the first thread.

17. The method of claim 16 wherein generating performance measurements for respective ones of the sub-threads of the first thread based at least in part on the cumulative suspend time of the first thread comprises:

taking a first timestamp to indicate start of a performance interval;

storing a starting value of the suspension number counter in conjunction with taking the first timestamp;

taking a second timestamp to indicate end of the performance interval;

storing an ending value of the suspension number counter in conjunction with taking the second timestamp;

utilizing the starting and ending values of the suspension number counter to determine respective corresponding values of the cumulative suspend time of the first thread; and

adjusting a difference between the first and second timestamps based at least in part on a difference between the corresponding values of the cumulative suspend time of the first thread.

18. A computer program product comprising a non-transitory processor-readable storage medium having stored therein program code of one or more software programs, wherein the program code when executed by at least one processing device comprising a processor coupled to a memory, causes said at least one processing device:

to monitor performance of a first thread of a first application executing on one of a plurality of processing cores of a storage system;

the first thread comprising an internal scheduler configured to control switching between a plurality of sub-threads of the first thread, and an external scheduler configured to control release of the processing core by the first thread for use by at least a second thread of a second application different than the first application;

in conjunction with monitoring the performance of the first thread in executing the first application, to maintain a cumulative suspend time of the first thread over multiple suspensions of the first thread, one or more of the multiple suspensions allowing at least the second thread of the second application to execute on the processing core; and

to generate performance measurements for respective ones of the sub-threads of the first thread based at least in part on the cumulative suspend time of the first thread.

19. The computer program product of claim 18 wherein maintaining a cumulative suspend time of the first thread over multiple suspensions of the first thread comprises, for each of the multiple suspensions of the first thread:

taking a first timestamp to indicate start of the suspension;

taking a second timestamp to indicate resumption of the first thread from the suspension; and

adding a time interval corresponding to a difference between the first and second timestamps to the cumulative suspend time.

20. The computer program product of claim 18 wherein maintaining a cumulative suspend time of the first thread over multiple suspensions of the first thread comprises, for each of the multiple suspensions of the first thread, storing a current value of the cumulative suspend time in association with a current value of a suspension number counter that is incremented with each of the multiple suspensions of the first thread.