Adaptive limited-duration edge resource management

1. A system for limited duration resource allocation in an edge computing network comprising:

at least one processor; and

memory including instructions that, when executed by the at least one processor, cause the at least one processor to perform operations to:

calculate available capacity for resources of a node of the edge computing network based on workloads executing on the node;

determine available set-aside resources of the node based on the available capacity, wherein the available set-aside resources designate resources of the node to be assigned to a workload for a limited duration of time;

receive a service request from an application executing on the node;

predict a first resource release rate for the workloads executing on the node using predicted completion times of the workloads;

determine a priority category for the service request based in-part on a second resource release rate of the application, wherein the second resource release rate is determined based on attestation parameters of the application; and

assign, for a limited duration of time, a set-aside resource from the available set-aside resources of the node to a workload associated with the service request based on the priority category based on the first resource release rate.

2. The system of claim 1 , wherein the priority category is best effort and the memory further comprising instructions that cause the at least one processor to perform operations to:

assign a preemption delay to the set-aside resources assigned to the workload;

receive a second service request with a higher priority category;

reclaim the set-aside resources from the workload; and

reassign the set-aside resources to a second workload associated with the second service request.

3. The system of claim 1 , the memory further comprising instructions that cause the at least one processor to perform operations to:

set a resource preemption schedule for the set-aside resources assigned to the workload based on the priority category;

determine that the priority category of the workload has changed; and

adjust the resource preemption schedule.

4. The system of claim 1 , the memory further comprising instructions that cause the at least one processor to perform operations to:

predict a future resource availability value for the node for a future time period; and

determine the set-aside resources based in part on the future resource availability value.

5. The system of claim 4 , the memory further comprising instructions that cause the at least one processor to perform operations to:

identify execution parameters of workload tasks executing on the node;

generate a distributed execution plan for the node using the execution parameters; and

predict the future resource availability using the distributed execution plan.

6. The system of claim 1 , the memory further comprising instructions that cause the at least one processor to perform operations to:

maintain a set-aside resource availability register for the set-aside resources; and

update the set-aside resource availability register based on the set-aside resources assigned to the workload.

7. The system of claim 6 , the memory further comprising instructions that cause the at least one processor to perform operations to:

determine that there are sufficient resources available at the node;

assign resources to the workload from available resources of the node;

reclaim the set-aside resources assigned to the workload; and

adjust the set-aside resource availability register based on the set-aside resources reclaimed from the workload.

8. The system of claim 1 , the memory further comprising instructions that cause the at least one processor to perform operations to:

collect telemetry data from the node; and

evaluate the telemetry data to estimate resource usage of workloads executing on the node, wherein the available set-aside resources is determined in part using the estimated resource usage.

9. At least one non-transitory machine-readable medium including instructions for limited duration resource allocation in an edge computing network that, when executed by at least one processor, cause the at least one processor to perform operations to:

calculate available capacity for resources of a node of the edge computing network based on workloads executing on the node;

determine available set-aside resources of the node based on the available capacity, wherein the available set-aside resources designate resources of the node to be assigned to a workload for a limited duration of time;

receive a service request from an application executing on the node;

predict a first resource release rate for the workloads executing on the node using predicted completion times of the workloads;

determine a priority category for the service request based in-part on a second resource release rate of the application, wherein the second resource release rate is determined based on attestation parameters of the application; and

assign, for a limited duration of time, a set-aside resource from the available set-aside resources of the node to a workload associated with the service request based on the priority category based on the first resource release rate.

10. The at least one non-transitory machine-readable medium of claim 9 , wherein the priority category is best effort and further comprising instructions that cause the at least one processor to perform operations to:

assign a preemption delay to the set-aside resources assigned to the workload;

receive a second service request with a higher priority category;

reclaim the set-aside resources from the workload; and

reassign the set-aside resources to a second workload associated with the second service request.

11. The at least one non-transitory machine-readable medium of claim 9 , further comprising instructions that cause the at least one processor to perform operations to:

set a resource preemption schedule for the set-aside resources assigned to the workload based on the priority category;

determine that the priority category of the workload has changed; and

adjust the resource preemption schedule.

12. The at least one non-transitory machine-readable medium of claim 9 , further comprising instructions that cause the at least one processor to perform operations to:

predict a future resource availability value for the node for a future time period; and

determine the set-aside resources based in part on the future resource availability value.

13. The at least one non-transitory machine-readable medium of claim 12 , further comprising instructions that cause the at least one processor to perform operations to:

identify execution parameters of workload tasks executing on the node;

generate a distributed execution plan for the node using the execution parameters; and

predict the future resource availability using the distributed execution plan.

14. The at least one non-transitory machine-readable medium of claim 9 , further comprising instructions that cause the at least one processor to perform operations to:

maintain a set-aside resource availability register for the set-aside resources; and

update the set-aside resource availability register based on the set-aside resources assigned to the workload.

15. The at least one non-transitory machine-readable medium of claim 14 , further comprising instructions that cause the at least one processor to perform operations to:

determine that there are sufficient resources available at the node;

assign resources to the workload from available resources of the node;

reclaim the set-aside resources assigned to the workload; and

adjust the set-aside resource availability register based on the set-aside resources reclaimed from the workload.

16. The at least one non-transitory machine-readable medium of claim 9 , further comprising instructions that cause the at least one processor to perform operations to:

collect telemetry data from the node; and

evaluate the telemetry data to estimate resource usage of workloads executing on the node, wherein the available set-aside resources is determined in part using the estimated resource usage.

17. A method for limited duration resource allocation in an edge computing network comprising:

calculating available capacity for resources of a node of the edge computing network based on workloads executing on the node;

determining available set-aside resources of the node based on the available capacity, wherein the available set-aside resources designate resources of the node to be assigned to a workload for a limited duration of time;

receiving a service request from an application executing on the node;

predicting a first resource release rate for the workloads executing on the node using predicted completion times of the workloads;

determining a priority category for the service request based in-part on a second resource release rate of the application, wherein the second resource release rate is determined based on attestation parameters of the application; and

assigning, for a limited duration of time, a set-aside resource from the available set-aside resources of the node to a workload associated with the service request based on the priority category based on the first resource release rate.

18. The method of claim 17 , wherein the priority category is best effort and further comprising:

assigning a preemption delay to the set-aside resources assigned to the workload;

receiving a second service request with a higher priority category;

reclaiming the set-aside resources from the workload; and

reassigning the set-aside resources to a second workload associated with the second service request.

19. The method of claim 17 , further comprising:

setting a resource preemption schedule for the set-aside resources assigned to the workload based on the priority category;

determining that the priority category of the workload has changed; and

adjusting the resource preemption schedule.

20. The method of claim 17 , further comprising:

predicting a future resource availability value for the node for a future time period; and

determining the set-aside resources based in part on the future resource availability value.

21. The method of claim 20 , further comprising:

identifying execution parameters of workload tasks executing on the node;

generating a distributed execution plan for the node using the execution parameters; and

predicting the future resource availability using the distributed execution plan.

22. The method of claim 17 , further comprising:

maintaining a set-aside resource availability register for the set-aside resources; and

updating the set-aside resource availability register based on the set-aside resources assigned to the workload.

23. The method of claim 22 , further comprising:

determining that there are sufficient resources available at the node;

assigning resources to the workload from available resources of the node;

reclaiming the set-aside resources assigned to the workload; and

adjusting the set-aside resource availability register based on the set-aside resources reclaimed from the workload.

24. The method of claim 17 , further comprising:

collecting telemetry data from the node; and

evaluating the telemetry data to estimate resource usage of workloads executing on the node, wherein the available set-aside resources is determined in part using the estimated resource usage.