On-demand code execution in a localized device coordinator

• US 10,452,439 B2
• Filed: 11/28/2016
• Issued: 10/22/2019
• Est. Priority Date: 11/28/2016
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Kuo, Calvin Yue-Ren, Amin, Rudresh, Chen, Yen-Lin, Turow, Jonathan I., Pan, Cheng, Brooker, Marc John, Sorenson, III, James Christopher, McClenahan, Weili Zhong
• Primary Examiner(s): Zhao, Bing
• Application Number:
  US15/362,696
• Publication Number:
  US 20180150325A1
• Time in Patent Office: 1,058 Days
• Field of Search: None
• CPC Class Codes: G06F 9/4843 by program, e.g. task dispatcher, supervisor, operating system
• CPC Class Codes: G06F 9/5005 to service a request
• CPC Class Codes: G06F 9/505 considering the load

Abstract

Systems and methods are described for implementing a coordinator within a coordinated environment, which environment includes set of coordinated devices managed by the coordinator. The coordinator can be provisioned with a set of tasks, each task corresponding to a segment of code that can be executed by the coordinator, such as to manage the coordinated devices. To assist in execution of the tasks, the coordinator can include a resource manager that generates execution environments corresponding to an individual task execution, which environments may be isolated from other execution environments. The coordinator can further include a scheduler to obtain calls to execute tasks, and implement a queue from which to select tasks for execution. On selecting a task, the scheduler may interact with the resource manager to generate an execution environment for the task, and cause execution of the task within the execution environment.

Claims

1. A coordinator computing device configured to manage one or more coordinated devices within a coordinated environment, wherein the one or more coordinated devices are distinct from the coordinator computing device, the coordinator computing device comprising:

a non-transitory data store including computer executable instructions corresponding to;

a scheduler that, when executed, implements a queue of calls to execute one or more tasks on the coordinator, wherein individual tasks of the one or more tasks correspond to code executable by the coordinator computing device to manage operation of the one or more coordinated devices within the coordinated environment;

a resource manager that, when executed, generates execution environments in which to execute the one or more tasks;

a processor in communication with the non-transitory data store and configured to execute the scheduler, wherein execution of the scheduler causes the processor to;

enqueue a set of task calls within the queue, individual task calls corresponding to a call to execute a task of the one or more tasks at the coordinator computing device;

obtain an availability of compute resources on the coordinator computing device from the resource manager;

apply a scheduling algorithm to the queue based at least in part on the availability of compute resources on the coordinator computing device to select at least one task call of the set of task calls for processing, wherein the coordinator computing device lacks sufficient compute resources to concurrently process both a task called by the selected at least one task call and tasks called by unselected task calls of the set of task calls, and wherein apply a scheduling algorithm to the queue comprises delaying processing on the coordinator computing device of the tasks called by the unselected task calls until sufficient compute resources become available on the coordinator computing device;

instruct the resource manager to select an execution environment on the coordinator computing device in which to execute the task corresponding to the at least one task call; and

execute the task corresponding to the at least one task call within the execution environment on the coordinator computing device, wherein the task, when executed, manages operation of at least one of the one or more coordinated devices within the coordinated environment;

wherein the processor is further configured to execute the resource manager in order to generate the execution environment on the coordinator computing device in which to execute the task corresponding to the at least one task call, and wherein generation of the execution environment includes provisioning the execution environment with a data resource required to execute the task corresponding to the at least one task call.

2. The system of claim 1, wherein the coordinated devices comprise at least one of a general purpose computing device, a thin device, or an embedded device associated with an alternative primary function.

3. The system of claim 1, wherein provisioning the execution environment with a data resource required to execute the task corresponding to the at least one task call includes provisioning the execution environment with the executable code corresponding to the task.

4. The system of claim 1, wherein provisioning the execution environment with a data resource required to execute the task corresponding to the at least one task call includes provisioning the execution environment with at least one of a driver, a code library, data corresponding to a runtime environment, or a code utility.

5. The system of claim 1, wherein the execution environment corresponds to a software container.

6. The system of claim 5, wherein the data resource is stored external to the software container, and wherein provisioning the execution environment with a data resource required to execute the task corresponding to the at least one task call includes associating the software container with read-only access to the data resource.

7. The system of claim 6, wherein associating the software container with read-only access to the data resource comprises generating a union mount combining read-only access to the data resource and read/write access to an additional memory space while presenting to the execution environment a combined view of the data resource and the additional memory space.

8. The system of claim 1, wherein the execution environment is a first execution environment, and wherein execution of the resource manager further causes the processor to:

obtain a request for a second execution environment in which to execute a second task;

detect that the first task has ceased execution; and

select the first execution environment to fulfill the request.

9. Non-transitory computer readable media including computer-executable instructions comprising:

first instructions executable by a computing system that is configured to manage one or more coordinated devices within a coordinated environment, the first instructions executable to implement a resource manager, wherein implementation of the resource manager causes the computing system to;

obtain a request for an execution environment on the computing system in which to execute a first task of one or more tasks, wherein individual tasks of the one or more tasks correspond to code executable by the computing system to manage operation of the one or more coordinated devices within the coordinated environment;

determine a set of data resources required for execution of the task; and

generate an execution environment on the computing system for execution of the task at least in part by provisioning the execution environment with access to the set of data resources; and

second instructions executable by the computing system to implement scheduler, wherein implementation of the scheduler causes the computing system to;

enqueue a set of task calls within a queue, individual task calls corresponding to a call to execute a task of the one or more tasks at the computing system;

obtain an availability of compute resources on the computing system;

apply a scheduling algorithm to the queue based at least in part on the availability of compute resources on the computing system to select a task call of the set of task calls for processing, wherein the task call corresponds to a call to execute the first task, wherein the computing system lacks sufficient compute resources to concurrently process both a task called by the selected at least one task call and tasks called by unselected task calls of the set of task calls, and wherein applying the scheduling algorithm to the queue comprises delaying processing on the computing system of the tasks called by the unselected task calls until sufficient compute resources become available on the computing system;

transmit to the resource manager the request for the execution environment on the computing system in which to execute the first task; and

execute the first task within the execution environment on the computing system, wherein the first task, when executed, manages operation of at least one of the one or more coordinated devices within the coordinated environment.

10. The non-transitory computer-readable media of claim 9, wherein the execution environment is a first execution environment, and wherein implementation of the resource manager further causes the computing system to:

obtain a request for a second execution environment in which to execute a second task;

detect that the first task has ceased execution; and

select the first execution environment to fulfill the request.

11. The non-transitory computer-readable media of claim 10, wherein the first execution environment is selected based at least in part on detecting that the second task and the first task correspond to the same task.

12. The non-transitory computer-readable media of claim 10, wherein the first execution environment is selected based at least in part on detecting that the second task depends on a data resource available to the first execution environment.

13. The non-transitory computer-readable media of claim 10, wherein implementation of the resource manager further causes the computing system to:

determine an availability of a compute resource required by the execution environment is below a threshold level; and

suspend a second execution environment associated with usage of the compute resource.

14. The non-transitory computer-readable media of claim 13, wherein the compute resource includes at least one of processor availability, memory availability, or network bandwidth availability.

15. The non-transitory computer-readable media of claim 13, wherein suspension of the second execution environment includes at least one of generating snapshot image data of the second execution environment or saving a state of the second execution environment to a device shadow.

16. The non-transitory computer-readable media of claim 9, wherein implementation of the resource manager further causes the computing system to generate the execution environment, and wherein generating the execution environment comprises generating a union mount combining read-only access to the set of data resources and read/write access to an additional memory space while presenting to the execution environment a combined view of the set of data resources and the additional memory space.

17. A computer-implemented method implemented by a computing system configured to manage one or more coordinated devices within a coordinated environment, the computer-implemented method comprising:

enqueuing a set of task calls within a queue, individual task calls corresponding to a call to execute a task of the one or more tasks on the computing system, wherein individual tasks of the one or more tasks correspond to code executable by the computing system to manage operation of the one or more coordinated devices within the coordinated environment;

obtaining an availability of compute resources on the computing system;

applying a scheduling algorithm to the queue based at least in part on the availability of compute resources on the computing system to select a task call of the set of task calls for processing, wherein the task call corresponds to a call to execute the first task, wherein the computing system lacks sufficient compute resources to concurrently process both a task called by the selected at least one task call and tasks called by unselected task calls of the set of task calls, and wherein applying the scheduling algorithm to the queue comprises delaying processing on the computing system of the tasks called by the unselected task calls until sufficient compute resources become available on the computing system;

determining a set of data resources required for execution of the task on the computing system;

generating an execution environment on the computing system for execution of the task at least in part by provisioning the execution environment with access to the set of data resources; and

executing the first task within the execution environment on the computing system, wherein the first task, when executed, manages operation of at least one of the one or more coordinated devices within the coordinated environment.

18. The computer-implemented method of claim 17, wherein the execution environment corresponds to a software container, and wherein provisioning the execution environment with access to the set of data resources comprises associating the software container with read-only access to the set of data resources.

19. The computer-implemented method of claim 17, wherein associating the software container with read-only access to the set of data resources comprises generating a union mount combining read-only access to the set of data resources and read/write access to an additional memory space while presenting to the execution environment a combined view of the set of data resources and the additional memory space.

20. The computer-implemented method of claim 17, wherein the execution environment is a first execution environment, and wherein the computer-implemented method further comprises:

obtaining a request for a second execution environment in which to execute a second task;

detecting that the first task has ceased execution; and

selecting the first execution environment to fulfill the request.

21. The computer-implemented method of claim 20, wherein the first execution environment is selected based at least in part on detecting at least one of that the second task and the first task correspond to the same task or that the second task depends on a data resource available to the first execution environment.

22. The computer-implemented method of claim 17 further comprising:

determining an availability of a compute resource required by the execution environment is below a threshold level; and

suspending a second execution environment associated with usage of the compute resource.

23. The computer-implemented method of claim 22, wherein suspending the second execution environment includes at least one of generating snapshot image data of the second execution environment or saving a state of the second execution environment to a device shadow.

24. The computer-implemented method of claim 22 further comprising:

enqueing a call to resume an execution the second execution environment within the queue;

processing the queue based at least in part on an availability of the compute resources to select the call to resume the execution of the second execution environment; and

resuming the execution of the second execution environment.