CARBON-AWARE INTELLIGENT POWER MANAGER FOR CLUSTER NODES

1 . A computing device comprising:

one or more memories; and

one or more processors communicatively coupled to the one or more memories, the one or more processors being configured to:

determine, based on executing at least one machine learning model, a measure of node criticality for a node of a cluster;

determine, based on the measure of node criticality, a power savings measure of one or more power savings measures to be applied to the node; and

apply the power savings measure to the node.

2 . The computing device of claim 1 , wherein the measure of node criticality comprises a weight indicative of an impact on at least one of scalability or availability caused by taking one of the one or more power savings measures.

3 . The computing device of claim 1 , wherein the one or more processors are configured to determine a corresponding measure of node criticality for each node of the cluster.

4 . The computing device of claim 1 , wherein the one or more processors are configured to determine the measure of node criticality on a periodic basis.

5 . The computing device of claim 1 , wherein to determine the measure of node criticality, the one or more processors are configured to determine a measure of a total number of predicted workloads to be scheduled on the node during a prediction window.

6 . The computing device of claim 5 , wherein the measure of the total number of predicted workloads to be scheduled on the node comprises a node scalability dependency factor indicative of a number of workloads predicted to be scheduled on the node during the prediction window due to scale demands and a node availability dependency factor indicative of a number of workloads predicted to be scheduled on the node during the prediction window due to availability demands.

7 . The computing device of claim 6 , wherein the node scalability dependency factor comprises the number of workloads predicted to be scheduled on the node during the prediction window due to scale demands divided by a total number of workloads for the cluster, and wherein the node availability dependency factor comprises the number of workloads predicted to be scheduled on the node during the prediction window due to availability demands divided by the total number of workloads for the cluster.

8 . The computing device of claim 1 , wherein to determine the measure of node criticality, the one or more processors are configured to:

determine a measure of node carbon emission for the node, the measure of node carbon emission being indicative of an amount of carbon emission per a unit of time attributable to the node; and

determine a measure of node resource utilization for the node, the measure of node resource utilization being indicative of a percentage of node resources that are utilized.

9 . The computing device of claim 1 , wherein the one or more processors are further configured to:

prior to determining the measure of node criticality, determine a spare node count, wherein the spare node count is indicative of a number of nodes of the cluster that are not necessary to meet predicted workloads of the cluster, and

determine that the spare node count is greater than zero, wherein determining the measure of node criticality is based on the determination that the spare node count is greater than zero.

10 . The computing device of claim 1 , wherein to determine the measure of node criticality, the one or more processors are configured to determine at least one of scalability or availability metrics, the scalability or availability metrics comprising at least one of:

a server utilization of the node, the server utilization of the node comprising a utilization percentage of node resources against a capacity of the node;

a carbon emission rate of node, the carbon emission rate of the node comprising an indication of an amount of carbon emission attributed to the node per a unit of time;

a workload scale factor, the workload scale factor being indicative of a number of service replicas predicted to be spawned in a prediction window; or

a workload availability factor, the workload availability factor being indicative of a number of standby replicas predicted to be spawned in the prediction window.

11 . The computing device of claim 10 , wherein to determine the measure of node criticality, the one or more processors are configured to, based on the at least one of scalability or availability metrics, determine at least one node dependency factor metric, the at least one node dependency factor metric comprising:

a node scalability dependency factor, the node scalability dependency factor being indicative of a number of workloads that are predicted to be scheduled on the node in the prediction window to achieve respective scalability goals associated with the workloads; or

a node availability dependency factor, the node availability dependency factor being indicative of a number of workloads that are predicted to be scheduled on the node in the prediction window to achieve respective availability goals associated with the workloads.

12 . The computing device of claim 1 , wherein the one or more power savings measures comprise at least one of:

shutting down at least one network interface card of the node;

lowering a frequency of at least one processor core of the node; or

moving the at least one processor core of the node into a different power optimized state.

13 . A method comprising:

determining, by one or more processors and based on executing of at least one machine learning model, a measure of node criticality for a node of a cluster;

determining, by the one or more processors and based on the measure of node criticality, a power savings measure of one or more power savings measures to be applied to the node; and

applying, by the one or more processors, the power savings measure to the node.

14 . The method of claim 13 , wherein the measure of node criticality comprises a weight indicative of an impact on at least one of scalability or availability caused by taking one of the one or more power savings measures.

15 . The method of claim 13 , wherein the method comprises determining, by the one or more processors, corresponding measure of node criticality for each node of a cluster.

16 . The method of claim 13 , wherein determining the measure of node criticality comprises determining a measure of a total number of predicted workloads to be scheduled on the node during a prediction window.

17 . The method of claim 13 , wherein determining the measure of node criticality comprises:

determining a measure of node carbon emission for the node, the measure of node carbon emission being indicative of an amount of carbon emission per a unit of time attributable to the node; and

determining a measure of node resource utilization for the node, the measure of node resource utilization being indicative of a percentage of node resources that are utilized.

18 . The method of claim 13 , further comprising:

determining, by the one or more processors and prior to determining the measure of node criticality, a spare node count, wherein the spare node count is indicative of a number of nodes of the cluster that are not necessary to meet predicted workloads of the cluster, and determining by the one or more processors, that the spare node count is greater than zero,

wherein determining the measure of node criticality is based on the determination that the spare node count is greater than zero.

19 . The method of claim 13 , wherein the one or more power savings measures comprise at least one of:

shutting down at least one network interface card of the node;

lowering a frequency of at least one processor core of the node; or

moving the at least one processor core of the node into a different power optimized state.

20 . Non-transitory computer-readable media, storing instructions which, when executed, cause one or more processors to:

determine, based on executing of at least one machine learning model, a measure of node criticality for a node of a cluster;

determine, based on the measure of node criticality, a power savings measure of one or more power savings measures to be applied to the node; and

apply the power savings measure to the node.