Granted Patent B2

US 11,886,911 · App. 16/915,380 · Granted Jan 30, 2024

Inventors: Vladimir Shveidel (Pardes-Hana, IL); Lior Kamran (Rishon LeZion, IL)

Assignee: EMC IP Holding Company LLC

G06F9/4881G06F3/067G06F3/0611G06F3/0659G06F9/5038G06F9/52

Abstract

Claims (92)

1. An apparatus comprising:

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

said at least one processing device being configured:

to associate different classes of service with respective threads of one or more applications executing on at least one of a plurality of processing cores of a storage system;

to configure different sets of prioritized thread queues for respective ones of the different classes of service;

to enqueue particular ones of the threads associated with particular ones of the classes of service in corresponding ones of the prioritized thread queues; and

to implement different dequeuing policies for selecting particular ones of the enqueued threads from the different sets of prioritized thread queues based at least in part on the different classes of service;

wherein the threads of the one or more applications executing on at least one of a plurality of processing cores of the storage system comprise different sub-threads of a first thread of a first application executing on a particular one of the plurality of processing cores of the storage system;

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

wherein different ones of the sub-threads that are in the suspended state are enqueued in respective different ones of the prioritized thread queues in order to wait for access to respective corresponding synchronization objects associated with resources of the particular processing core.

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 and the first thread comprises a truck thread that when executing on the particular processing core is configured to utilize substantially all available processing resources of that processing core.

4. The apparatus of claim 1 wherein associating different classes of service with respective threads of one or more applications comprises assigning designated class tags as attributes of the respective threads wherein different ones of the class tags denote respective different ones of the classes of service.

5. The apparatus of claim 4 wherein said at least one processing device is further configured:

to receive an input-output operation for processing;

to assign a particular one of the class tags to the input-output operation; and

to assign the particular one of the class tags to each of a plurality of threads that are generated in conjunction with the processing of the input-output operation;

wherein the assignment of the particular one of the class tags to the input-output operation is persistent throughout the processing of the input-output operation.

6. The apparatus of claim 5 wherein the processing of the input-output operation is distributed across a plurality of distinct storage nodes of the storage system, and further wherein each such storage node comprises one or more of the plurality of processing cores of the storage system.

7. The apparatus of claim 6 wherein the assigned class tag for the input-output operation is passed from a first one of the storage nodes to a second one of the storage nodes in conjunction with the processing of the input-output operation and is assigned to one or more threads generated in the second storage node for the processing of the input-output operation.

8. The apparatus of claim 1 wherein uniform sets of prioritized thread queues are provided for respective ones of a plurality of different synchronization objects, each of the synchronization objects comprising one or more of a semaphore, a timer, a lock, a barrier, a memory pool and a thread pool, wherein a single non-prioritized thread queue of a given one of the synchronization objects is replaced with multiple thread queues and priority-based selection among the multiple thread queues for the given synchronization object.

9. An apparatus comprising:

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

said at least one processing device being configured:

to associate different classes of service with respective threads of one or more applications executing on at least one of a plurality of processing cores of a storage system;

to configure different sets of prioritized thread queues for respective ones of the different classes of service;

to enqueue particular ones of the threads associated with particular ones of the classes of service in corresponding ones of the prioritized thread queues; and

to implement different dequeuing policies for selecting particular ones of the enqueued threads from the different sets of prioritized thread queues based at least in part on the different classes of service;

wherein the different sets of prioritized thread queues configured for respective ones of the different classes of service provide different dynamically adjustable latency levels for the threads that are enqueued therein.

10. An apparatus comprising:

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

said at least one processing device being configured:

to associate different classes of service with respective threads of one or more applications executing on at least one of a plurality of processing cores of a storage system;

to configure different sets of prioritized thread queues for respective ones of the different classes of service;

to enqueue particular ones of the threads associated with particular ones of the classes of service in corresponding ones of the prioritized thread queues; and

to implement different dequeuing policies for selecting particular ones of the enqueued threads from the different sets of prioritized thread queues based at least in part on the different classes of service;

wherein configuring different sets of prioritized thread queues for respective ones of the different classes of service comprises:

determining a number of priority classes;

determining a number of selection indexes for respective ones of the priority classes; and

establishing a particular number of prioritized thread queues based at least in part on the number of priority classes and the number of selection indexes.

11. An apparatus comprising:

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

said at least one processing device being configured:

to associate different classes of service with respective threads of one or more applications executing on at least one of a plurality of processing cores of a storage system;

to configure different sets of prioritized thread queues for respective ones of the different classes of service;

to enqueue particular ones of the threads associated with particular ones of the classes of service in corresponding ones of the prioritized thread queues; and

to implement different dequeuing policies for selecting particular ones of the enqueued threads from the different sets of prioritized thread queues based at least in part on the different classes of service;

wherein configuring different sets of prioritized thread queues for respective ones of the different classes of service comprises:

determining a number of priority classes;

establishing a set of prioritized thread queues for each of the priority classes; and

assigning weights to respective ones of the sets of prioritized thread queues of the respective priority classes;

wherein the prioritized thread queues within a given one of the sets each have a same priority for selection of threads enqueued therein; and

wherein different ones of the prioritized thread queues in different ones of the sets have different priorities for selection of threads enqueued therein.

12. 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 associate different classes of service with respective threads of one or more applications executing on at least one of a plurality of processing cores of a storage system;

to configure different sets of prioritized thread queues for respective ones of the different classes of service;

to enqueue particular ones of the threads associated with particular ones of the classes of service in corresponding ones of the prioritized thread queues; and

to implement different dequeuing policies for selecting particular ones of the enqueued threads from the different sets of prioritized thread queues based at least in part on the different classes of service;

wherein the threads of the one or more applications executing on at least one of a plurality of processing cores of the storage system comprise different sub-threads of a first thread of a first application executing on a particular one of the plurality of processing cores of the storage system;

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

wherein different ones of the sub-threads that are in the suspended state are enqueued in respective different ones of the prioritized thread queues in order to wait for access to respective corresponding synchronization objects associated with resources of the particular processing core.

13. The computer program product of claim 12 wherein associating different classes of service with respective threads of one or more applications comprises assigning designated class tags as attributes of the respective threads wherein different ones of the class tags denote respective different ones of the classes of service.

14. The computer program product of claim 12 wherein the different sets of prioritized thread queues configured for respective ones of the different classes of service provide different dynamically adjustable latency levels for the threads that are enqueued therein.

15. A method comprising:

associating different classes of service with respective threads of one or more applications executing on at least one of a plurality of processing cores of a storage system;

configuring different sets of prioritized thread queues for respective ones of the different classes of service;

enqueuing particular ones of the threads associated with particular ones of the classes of service in corresponding ones of the prioritized thread queues; and

implementing different dequeuing policies for selecting particular ones of the enqueued threads from the different sets of prioritized thread queues based at least in part on the different classes of service;

wherein the threads of the one or more applications executing on at least one of a plurality of processing cores of the storage system comprise different sub-threads of a first thread of a first application executing on a particular one of the plurality of processing cores of the storage system;

wherein each of the sub-threads of the first thread is in a particular one of multiple designated states at a particular point in time, the multiple designated states including a running state, a ready state and a suspended state; and

wherein different ones of the sub-threads that are in the suspended state are enqueued in respective different ones of the prioritized thread queues in order to wait for access to respective corresponding synchronization objects associated with resources of the particular processing core.

16. The method of claim 15 wherein associating different classes of service with respective threads of one or more applications comprises assigning designated class tags as attributes of the respective threads wherein different ones of the class tags denote respective different ones of the classes of service.

17. The method of claim 15 wherein the different sets of prioritized thread queues configured for respective ones of the different classes of service provide different dynamically adjustable latency levels for the threads that are enqueued therein.

18. The method of claim 15 wherein configuring different sets of prioritized thread queues for respective ones of the different classes of service comprises:

determining a number of priority classes;

establishing a set of prioritized thread queues for each of the priority classes; and

assigning weights to respective ones of the sets of prioritized thread queues of the respective priority classes;

wherein the prioritized thread queues within a given one of the sets each have a same priority for selection of threads enqueued therein; and

wherein different ones of the prioritized thread queues in different ones of the sets have different priorities for selection of threads enqueued therein.

19. The method of claim 15 wherein uniform sets of prioritized thread queues are provided for respective ones of a plurality of different synchronization objects, each of the synchronization objects comprising one or more of a semaphore, a timer, a lock, a barrier, a memory pool and a thread pool, wherein a single non-prioritized thread queue of a given one of the synchronization objects is replaced with multiple thread queues and priority-based selection among the multiple thread queues for the given synchronization object.

20. An apparatus comprising:

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

said at least one processing device being configured:

to associate different classes of service with respective threads of one or more applications executing on at least one of a plurality of processing cores of a storage system;

to configure different sets of prioritized thread queues for respective ones of the different classes of service;

to enqueue particular ones of the threads associated with particular ones of the classes of service in corresponding ones of the prioritized thread queues; and

to implement different dequeuing policies for selecting particular ones of the enqueued threads from the different sets of prioritized thread queues based at least in part on the different classes of service;

wherein associating different classes of service with respective threads of one or more applications comprises assigning designated class tags as attributes of the respective threads wherein different ones of the class tags denote respective different ones of the classes of service; and

wherein said at least one processing device is further configured:

to receive an input-output operation for processing;

to assign a particular one of the class tags to the input-output operation; and

to assign the particular one of the class tags to each of a plurality of threads that are generated in conjunction with the processing of the input-output operation.

Assignments (9)