Dynamically augmenting edge resources
Systems and methods may be used to determine where to run a service based on workload-based conditions or system-level conditions. An example method may include determining whether power available to a resource of a compute device satisfies a target power, for example to satisfy a target performance for a workload. When the power available is insufficient, an additional resource may be provided, for example on a remote device from the compute device. The additional resource may be used as a replacement for the resource of the compute device or to augment the resource of the compute device.
1. A compute device operable in an edge network, the compute device including a local resource pool to execute a workload in the edge network, the compute device comprising:
processing circuitry to:
receive execution conditions for the workload;
access capacity of the local resource pool, the capacity including available power supply to the local resource pool;
determine whether the capacity including the available power supply is sufficient to execute the workload based on the execution conditions for the workload;
provide, to an orchestrator logic unit in response to a determination that the available power is insufficient to execute the workload, a request to augment the local resource pool with resources of a remote resource pool that is independently powered;
receive, from the orchestrator logic unit, information indicative of an additional resource of the remote resource pool within the network outside the compute device; and
cause a first portion of the workload to be executed using the additional resource, wherein the available power supply is sufficient to execute a second portion of the workload after removing the first portion of the workload.
2. The compute device of claim 1 , wherein the workload is executed entirely using the additional resource without using the local resource pool.
3. The compute device of claim 1 , wherein the compute device is a node of a distributed autonomously powered system and the additional resource is on a remote server.
4. The compute device of claim 1 , wherein the processing circuitry is further configured to:
receive, from the orchestrator logic unit, connection data usable to communicatively couple with the additional resource to operate within the local resource pool of the compute device; and
communicatively couple, using the connection data, the compute device with the additional resource;
wherein to cause the first portion of the workload to be executed, the processing circuitry is configured to execute the workload using the local resource pool including the additional resource.
5. The compute device of claim 4 , wherein to communicatively couple the compute device with the additional resource, the processing circuitry is further configured to communicatively couple the compute device with at least one of a compute resource, an accelerator resource, or a data storage resource.
6. The compute device of claim 1 , wherein the processing circuitry is further configured to provide, to the orchestrator logic, system-level conditions for the local resource pool.
7. The compute device of claim 1 , wherein the workload-based conditions include at least one of latency requirements of the workload, a service level agreement related to the workload, a quality of service (QoS) agreement related to the workload, a time to complete the workload, power requirements for the workload, or location requirements for the workload.
8. The compute device of claim 1 , wherein the processing circuitry is further configured to execute a remaining portion of the workload using the local resource pool.
9. A method of configuring a compute device including a local resource pool to execute a workload in an edge network, the method comprising:
receiving execution conditions for the workload;
accessing capacity of the local resource pool, the capacity including available supply of power to the local resource pool;
determining, using a processor, whether the capacity including the available supply of power is sufficient to execute the workload based on the execution conditions for the workload;
providing, to an orchestrator logic unit in response to a determination that the available supply of power is insufficient to execute the workload, a request to augment the local resource pool with resources of a remote resource pool that is independently powered;
receiving, from the orchestrator logic unit, information indicative of an additional resource of the remote resource pool within the network outside the compute device; and
causing a first portion of the workload to be executed using the additional resource, wherein the available supply of power is sufficient to execute a second portion of the workload after removing the first portion of the workload.
10. The method of claim 9 , wherein the workload is executed entirely using the additional resource without using the local resource pool.
11. The method of claim 9 , wherein the compute device is a node of a distributed autonomously powered system and the additional resource is on a remote server.
12. The method of claim 9 , further comprising:
receiving, from the orchestrator logic unit, connection data usable to communicatively couple with the additional resource to operate within the local resource pool of the compute device; and
communicatively coupling, using the connection data, the compute device with the additional resource; and
wherein causing the first portion of the workload to be executed includes executing the workload using the local resource pool including the additional resource.
13. The method of claim 12 , wherein communicatively coupling the compute device with the additional resource includes communicatively coupling the compute device with at least one of a compute resource, an accelerator resource, or a data storage resource.
14. The method of claim 9 , further comprising providing, to the orchestrator logic, system-level conditions for the local resource pool.
15. The method of claim 9 , wherein the workload-based conditions include at least one of latency requirements of the workload, a service level agreement related to the workload, a quality of service (QoS) agreement related to the workload, a time to complete the workload, power requirements for the workload, or location requirements for the workload.
16. The method of claim 9 , further comprising executing a remaining portion of the workload using the local resource pool.
17. At least one non-transitory machine-readable medium including instructions for configuring a compute device including a local resource pool to execute a workload in an edge network, the instructions, when executed, causing processing circuitry to:
receive execution conditions for the workload;
access capacity of the local resource pool, the capacity including available power supply to the local resource pool;
determine whether the capacity including the available power supply is sufficient to execute the workload based on the execution conditions for the workload;
provide, to an orchestrator logic unit in response to a determination that the available power supply is insufficient to execute the workload, a request to augment the local resource pool with resources of a remote resource pool that is independently powered;
receive, from the orchestrator logic unit, information indicative of an additional resource of the remote resource pool within the network outside the compute device; and
cause a first portion of the workload to be executed using the additional resource, wherein the available power supply is sufficient to execute a second portion of the workload after removing the first portion of the workload.
18. The machine-readable medium of claim 17 , wherein the workload is executed entirely using the additional resource without using the local resource pool.
19. The machine-readable medium of claim 17 , wherein the compute device is a node of a distributed autonomously powered system and the additional resource is on a remote server.
20. The machine-readable medium of claim 17 , wherein the instructions further cause the processing circuitry to:
receive, from the orchestrator logic unit, connection data usable to communicatively couple with the additional resource to operate within the local resource pool of the compute device; and
communicatively couple, using the connection data, the compute device with the additional resource; and
wherein to cause the first portion of the workload to be executed, the instructions further cause the processing circuitry to execute the workload using the local resource pool including the additional resource.
21. The machine-readable medium of claim 20 , wherein to communicatively couple the compute device with the additional resource, the instructions further cause the processing circuitry to communicatively couple the compute device with at least one of a compute resource, an accelerator resource, or a data storage resource.
22. The machine-readable medium of claim 17 , wherein the instructions further cause the processing circuitry to provide, to the orchestrator logic, system-level conditions for the local resource pool.
23. The machine-readable medium of claim 17 , wherein the workload-based conditions include at least one of latency requirements of the workload, a service level agreement related to the workload, a quality of service (QoS) agreement related to the workload, a time to complete the workload, power requirements for the workload, or location requirements for the workload.
24. The machine-readable medium of claim 17 , wherein the instructions further cause the processing circuitry to execute a remaining portion of the workload using the local resource pool.