Policy-Based Dynamic Compute Unit Adjustments

1 . A method, comprising:

presenting a user interface indicating a plurality of policies specifying operational triggers and responsive actions for altering composition of compute units each comprising a plurality of physical computing components;

receiving a first user selection indicating a set of physical computing components to form a target compute unit;

receiving a second user selection indicating a selected policy among the plurality of policies to apply to the target compute unit;

instructing a management entity to establish the target compute unit based at least on logical partitioning within a communication fabric communicatively coupling the set of physical computing components; and

instructing the management entity to monitor telemetry data for the target compute unit and alter composition of the target compute unit using the logical partitioning responsive to at least one triggers indicated by the selected policy.

2 . The method of claim 1 , wherein the plurality of policies each comprise the operational triggers selected from among performance triggers, error triggers, and time triggers, and wherein responsive to meeting criteria specified for the operational triggers, the plurality of policies indicate to the management entity to add or remove one or more physical computing components from corresponding compute units.

3 . The method of claim 1 , further comprising:

instructing the management entity to deploy one or more telemetry elements to the target compute unit, wherein the one or more telemetry elements monitor operational properties of the target compute unit and provide the telemetry data to the management entity.

4 . The method of claim 1 , further comprising:

in the user interface:

presenting an option for creation of a new policy;

presenting indications of one or more triggers and one or more actions responsive to the one or more triggers for inclusion in the new policy;

receiving user selections among the one or more triggers and the one or more actions for inclusion in the new policy; and

storing a specification of the new policy for subsequent usage in adjusting composition of a compute unit after deployment of the compute unit.

5 . The method of claim 1 , further comprising:

altering the composition of the target compute unit by at least changing the logical partitioning among the set of physical computing components to add or remove at least one among the set of physical computing components from the compute unit, and rebooting a processor component remaining in the set of physical computing components.

6 . The method of claim 1 , wherein the plurality of physical computing components are selected from among central processing units (CPUs), graphics processing units (GPUs), data storage devices, field-programmable gate arrays (FPGAs), and network interface modules.

7 . An apparatus, comprising:

one or more computer readable storage media;

a processing system operatively coupled with the one or more computer readable storage media; and

program instructions stored on the one or more computer readable storage media that, based on being read and executed by the processing system, direct the processing system to at least:

present a user interface indicating a plurality of policies specifying operational triggers and responsive actions for altering composition of compute units each comprising a plurality of physical computing components;

receive a first user selection indicating a set of physical computing components to form a target compute unit;

receive a second user selection indicating a selected policy among the plurality of policies to apply to the target compute unit;

instruct a management entity to establish the target compute unit based at least on logical partitioning within a communication fabric communicatively coupling the set of physical computing components; and

instruct the management entity to monitor telemetry data for the target compute unit and alter composition of the target compute unit using the logical partitioning responsive to at least one trigger indicated by the selected policy.

8 . The apparatus of claim 7 , wherein the plurality of policies each comprise the operational triggers selected from among performance triggers, error triggers, and time triggers, and wherein responsive to meeting criteria specified for the operational triggers, the plurality of policies indicate to the management entity to add or remove one or more physical computing components from corresponding compute units.

9 . The apparatus of claim 7 , comprising further program instructions, based on being executed by the processing system, direct the processing system to at least:

instruct the management entity to deploy one or more telemetry elements to the target compute unit, wherein the one or more telemetry elements monitor operational properties of the target compute unit and provide the telemetry data to the management entity.

10 . The apparatus of claim 7 , comprising further program instructions, based on being executed by the processing system, direct the processing system to at least:

in the user interface:

present an option for creation of a new policy;

present indications of one or more triggers and one or more actions responsive to the triggers for inclusion in the new policy;

receive user selections among the one or more triggers and the one or more actions for inclusion in the new policy; and

store a specification of the new policy for subsequent usage in adjusting composition of a compute unit after deployment of the compute unit.

11 . The apparatus of claim 7 , comprising further program instructions, based on being executed by the processing system, direct the processing system to at least:

alter the composition of the target compute unit by at least changing the logical partitioning among the set of physical computing components to add or remove at least one among the set of physical computing components from the compute unit, and rebooting a processor component remaining in the set of physical computing components.

12 . The apparatus of claim 7 , wherein the plurality of physical computing components are selected from among central processing units (CPUs), graphics processing units (GPUs), data storage devices, field-programmable gate arrays (FPGAs), and network interface modules.

13 . A system, comprising:

a management processor configured to receive user commands to establish compute units among a plurality of physical computing components, each of the compute units comprising one or more of the plurality of physical computing components with at least one among the plurality of physical computing components configured to report telemetry data to the management processor related to operation of an associated compute unit;

a communication fabric configured to communicatively couple the plurality of physical computing components and form the compute units using logical partitioning within the communication fabric; and

the management processor configured to alter composition of the plurality of physical computing components within a target compute unit after formation of the target compute unit by at least changing the logical partitioning within the communication fabric responsive to corresponding telemetry data and a selected operational policy for the target compute unit.

14 . The system of claim 13 , comprising:

the management processor configured to present a user interface to receive the user commands, wherein the user interface indicates a plurality of policies specifying operational triggers and responsive actions for dynamically altering composition of the compute units.

15 . The system of claim 13 , comprising:

the management processor configured to alter the composition of the target compute unit by at least changing the logical partitioning among the set of physical computing components to add or remove at least one among the set of physical computing components from the compute unit, and rebooting a processor component remaining in the set of physical computing components.

16 . The system of claim 13 , comprising:

the management processor configured to receive a first user selection indicating a set of physical computing components to form the target compute unit; and

the management processor configured to receive a second user selection indicating the selected policy operational among a plurality of operational policies to apply to the target compute unit.

17 . The system of claim 16 , wherein the plurality of policies each comprise operational triggers selected from among performance triggers, error triggers, and time triggers, and wherein responsive to meeting criteria specified for the operational triggers, the plurality of operational policies indicate to the management processor to add or remove one or more physical computing components from the target compute unit.

18 . The system of claim 13 , comprising:

the management processor configured to deploy one or more telemetry elements to the target compute unit, wherein the one or more telemetry elements monitor operational properties of the target compute unit and provide the telemetry data to the management processor.

19 . The system of claim 13 , comprising:

the management processor configured to present a user interface configured to:

present an option for creation of a new policy;

present indications of one or more triggers and one or more actions responsive to the triggers for inclusion in the new policy;

receive user selections among the one or more triggers and the one or more actions for inclusion in the new policy; and

store a specification of the new policy for subsequent usage in adjusting composition of a compute unit after deployment of the compute unit.

20 . The system of claim 13 , wherein the plurality of physical computing components are selected from among central processing units (CPUs), graphics processing units (GPUs), data storage devices, field-programmable gate arrays (FPGAs), and network interface modules.