Method, Apparatus and System for Coordinating Execution of Tasks in a Computing System Having a Distributed Shared Memory

• US 20140373026A1
• Filed: 08/28/2014
• Published: 12/18/2014
• Est. Priority Date: 02/28/2012
• Status: Active Grant

• Current Assignee: Huawei Technologies Co., Ltd. (Huawei Investment & Holding Co., Ltd.)
• Original Assignee: Huawei Technologies Co., Ltd. (Huawei Investment & Holding Co., Ltd.)
• Inventors: Gu, Lin, Ma, Zhiqiang, Zeng, Yulong
• Granted Patent: US 9,201,691 B2
• Time in Patent Office: Days
• US Class Current: 718/104
• CPC Class Codes: G06F 2003/0697 device management, e.g. handlers, drivers, I/O schedulers
• CPC Class Codes: G06F 3/0604 Improving or facilitating administration, e.g. storage management
• CPC Class Codes: G06F 3/0619 in relation to data integrity, e.g. data losses, bit errors
• CPC Class Codes: G06F 3/0638 Organizing or formatting or addressing of data
• CPC Class Codes: G06F 3/065 Replication mechanisms
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]
• CPC Class Codes: G06F 9/4881 Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
• CPC Class Codes: G06F 9/5027 the resource being a machine, e.g. CPUs, Servers, Terminals
• CPC Class Codes: G06F 9/52 Program synchronisation; Mutual exclusion, e.g. by means of semaphores

Abstract

A task coordination apparatus in a computing system having a distributed shared memory (DSM) coordinates the execution of two related tasks, wherein the second task has an execution variable which is modified by the first task. The task coordination apparatus creates a snapshot of a memory space in the distributed shared memory assigned to the first task and a cooperation watching area of the second task. The cooperation watching area contains a memory address pointing to a location where the execution variable of the second task is stored in the memory space assigned to the first task. The first task is allocated to a first computing node for execution, and the memory space assigned to it is updated according to the execution result. After updating the memory space, the second task is allocated to a second computing node for execution using the execution variable updated by the first task.

Claims

1. A method for coordinating execution of tasks in a computing system having a distributed shared memory, comprising:

creating a snapshot of a memory space assigned to a first task, the memory space being a part of the distributed shared memory;

creating a cooperation watching area of a second task, wherein the second task has an execution variable that is modified by the first task, and wherein the cooperation watching area contains a memory address pointing to a location where the execution variable of the second task is stored in the memory space assigned to the first task;

allocating the first task to a first computing node of the computing system for execution;

updating, according to an execution result of the first task, the snapshot of the memory space and data in the memory space; and

allocating the second task to a second computing node of the computing system, including providing access to the memory address for the execution variable, for the second computing node to execute the second task using the execution variable after updating the memory space of the first task.

2. The method according to claim 1, wherein after creating the cooperation watching area of the second task, the method further comprises putting the second task into a watcher queue, and wherein the watcher queue is a queue storing a task that has a cooperation watching area but cannot be executed immediately.

3. The method according to claim 2, wherein the step of allocating the second task to the second computing node comprises moving the second task from the watcher queue to an execution queue, and wherein the execution queue is a queue storing a task to be executed immediately.

4. The method according to claim 3, wherein the step of moving the second task from the watcher queue to the execution queue comprises moving all tasks whose execution variables are updated by the first task from the watcher queue to the execution queue, and wherein all tasks are in the watcher queue and include the second task.

5. The method according to claim 1, wherein after the step of allocating the second task to the second computing node, the method further comprises:

determining, by the second computing node, whether the plurality of execution variables of the second task have been updated when a plurality of execution variables are needed for implementing the second task;

executing, by the second computing node, the second task based on a determination that the plurality of execution variables of the second task have been updated; and

putting the second task back into the watcher queue based on a determination that at least one of the plurality of execution variables of the second task has not been updated.

6. The method according to claim 1, wherein the cooperation watching area of the second task is located in execution context of the second task.

7. A task cooperation apparatus for coordinating execution of tasks in a computing system having a distributed shared memory, comprising a processor configured to:

create a snapshot of a memory space assigned to a first task, the memory space being a part of the distributed shared memory;

create a cooperation watching area of a second task, wherein the second task has an execution variable that is modified by the first task, and wherein the cooperation watching area contains a memory address pointing to a location where the execution variable of the second task is stored in the memory space assigned to the first task;

allocate the first task to a first computing node of the computing system for execution;

update, according to an execution result of the first task, the snapshot of the memory space and data in the memory space; and

allocate the second task to a second computing node of the computing system, including providing access to the memory address for the execution variable, for the second computing node to execute the second task using the execution variable after updating the memory space of the first task.

8. The apparatus according to claim 7, wherein the processor is further configured to put the second task into a watcher queue, and wherein the watcher queue is a queue storing a task that has a cooperation watching area but cannot be executed immediately.

9. The apparatus according to claim 8, wherein the operation of allocating the second task to the second computing node comprises move the second task from the watcher queue to an execution queue after the memory space of the first task is updated, and wherein the execution queue is a queue storing a task to be executed immediately.

10. The apparatus according to claim 9, wherein the processor is configured to move all tasks whose execution variables are updated by the first task from the watcher queue to the execution queue after updating the memory space of the first task, and wherein all tasks are in the watcher queue and comprise the second task.

11. The apparatus according to claim 7, wherein the cooperation watching area of the second task is located in execution context of the second task.

12. A computing system comprising:

a first computing node;

a second computing node;

a task cooperation apparatus; and

a distributed shared memory,

wherein the task cooperation apparatus is configured to;

create a snapshot of a memory space assigned to a first task, the memory space being a part of the distributed shared memory;

create a cooperation watching area of a second task, wherein the second task has an execution variable that is modified by the first task, and wherein the cooperation watching area contains a memory address pointing to a location where the execution variable of the second task is stored in the memory space assigned to the first task;

allocate the first task to a first computing node of the computing system for execution;

update, according to an execution result of the first task, the snapshot of the memory space and data in the memory space; and

allocate the second task to a second computing node of the computing system, including providing access to the memory address for the execution variable, for the second computing node to execute the second task using the execution variable after updating the memory space of the first task.

13. The computing system according to claim 12, wherein the task cooperation apparatus is further configured to put the second task into a watcher queue, and wherein the watcher queue is a queue storing a task that has a cooperation watching area but cannot be executed immediately.

14. The computing system according to claim 13, wherein the operation of allocating the second task to the second computing node by the task cooperation apparatus comprises moving the second task from the watcher queue to an execution queue after updating the memory space of the first task, and wherein the execution queue is a queue storing a task to be executed immediately.

15. The computing system according to claim 14, wherein the task cooperation apparatus is configured to move all tasks whose cooperation variables are updated by the first task from the watcher queue to the execution queue after the memory space of the first task is updated, and wherein the all tasks are in the watcher queue and comprise the second task.

16. The computing system according to claim 12, wherein a plurality of execution variables are used for implementing the second task, and wherein the second computing node is configured to:

determine whether the plurality of execution variables of the second task have been updated after the second task is allocated to the second computing node;

put the second task back into the watcher queue based on a determination that at least one of the plurality of execution variables of the second task has not been updated; and

execute the second task based on a determination that all memory spaces corresponding to the plurality of execution variables of the second task have been updated.

17. The computing system according to claim 12, wherein the task cooperation apparatus is located in the first computing node.

18. The computing system according to claim 12, wherein the task cooperation apparatus is located in the second computing node.

19. The computing system according to claim 12, wherein the task cooperation apparatus is connected to the first computing node and the second computing node.

20. The computing system according to claim 12, wherein the cooperation watching area of the second task is located in execution context of the second task.