Scheduling workloads in a container orchestrator of a virtualized computer system

Patent metadata

Field	Value
Publication number	US-12190140-B2
Application number	US-202117154776-A
Country	US
Kind code	B2
Filing date	Jan 21, 2021
Priority date	Jan 21, 2021
Publication date	Jan 7, 2025
Grant date	Jan 7, 2025

Abstract

An example method of scheduling a workload in a virtualized computing system including a host cluster having a virtualization layer directly executing on hardware platforms of hosts is described. The virtualization layer supports execution of virtual machines (VMs) and is integrated with an orchestration control plane. The method includes: receiving, at the orchestration control plane, a workload specification for the workload; selecting, at the orchestration control plane, a plurality of nodes for the workload based on the workload specification, each of the plurality of nodes implemented by a host of the hosts; selecting, by the orchestration control plane in cooperation with a virtualization management server managing the host cluster, a node of the plurality of nodes; and deploying, by the orchestration control plane in cooperation with the virtualization management server, the workload on a host in the host cluster implementing the selected node.

First claim

What is claimed is:

1. A method of scheduling a workload in a virtualized computing system including a host cluster having a virtualization layer directly executing on hardware platforms of hosts, the virtualization layer supporting execution of virtual machines (VMs), the virtualization layer integrated with an orchestration control plane, the method comprising:
receiving, at the orchestration control plane, a workload specification for the workload;
selecting, by a first scheduler of the orchestration control plane, a plurality of nodes as candidates for the workload based on the workload specification, each of the plurality of nodes implemented by a host of the hosts;
sending, by the first scheduler through a scheduler extender of the orchestration control plane, the plurality of nodes as candidate nodes to a second scheduler of a virtualization management server managing the host cluster;
selecting, by the second scheduler, a node of the plurality of nodes, the first scheduler being a slave the second scheduler, which is a master, the master receiving the candidates from the slave and selecting the node from the candidates on behalf of the slave;
receiving, at the first scheduler through the scheduler extender, the selected node from the second scheduler; and
deploying, by the orchestration control plane in cooperation with the virtualization management server, the workload on a host in the host cluster implementing the selected node, the orchestration control plane deploying the workload through communication with a controller of a hypervisor of the host, the controller managing a VM having an agent executing therein, the controller executing the workload in the VM through communication with the agent.

2. The method of claim 1 , wherein the VMs include pod VMs, the pod VMs including container engines supporting execution of containers in the pod VMs, wherein the workload is a pod and the workload specification is a pod specification, and wherein the pod specification is received at a master server of the orchestration control plane.

3. The method of claim 2 , wherein the pod specification includes first resource constraints for a node, and wherein the master server selects the plurality of nodes as candidate nodes based on the first resource constraints in the pod specification.

4. The method of claim 3 , wherein the virtualization management server converts the first resource constraints into second resource constraints for a host when selecting the node of the plurality of nodes.

5. The method of claim 4 , wherein the first resource constraints and the second resource constraints include constraints on at least one of processor, memory, network, and storage.

6. The method of claim 3 , further comprising:
converting, by the orchestration control plane, the pod specification to a VM specification for a pod VM.

7. A non-transitory computer readable medium comprising instructions to be executed in a computing device to cause the computing device to carry out a method of scheduling a workload in a virtualized computing system including a host cluster having a virtualization layer directly executing on hardware platforms of hosts, the virtualization layer supporting execution of virtual machines (VMs), the virtualization layer integrated with an orchestration control plane, the method comprising:
receiving, at the orchestration control plane, a workload specification for the workload;
selecting, by a first scheduler of the orchestration control plane, a plurality of nodes as candidates for the workload based on the workload specification, each of the plurality of nodes implemented by a host of the hosts;
sending, by the first scheduler through a scheduler extender of the orchestration control plane, the plurality of nodes as candidate nodes to a second scheduler of a virtualization management server managing the host cluster;
selecting, by the second scheduler, a node of the plurality of nodes, the first scheduler being a slave the second scheduler, which is a master, the master receiving the candidates from the slave and selecting the node from the candidates on behalf of the slave;
receiving, at the first scheduler through the scheduler extender, the selected node from the second scheduler; and
deploying, by the orchestration control plane in cooperation with the virtualization management server, the workload on a host in the host cluster implementing the selected node, the orchestration control plane deploying the workload through communication with a controller of a hypervisor of the host, the controller managing a VM having an agent executing therein, the controller executing the workload in the VM through communication with the agent.

8. The non-transitory computer readable medium of claim 7 , wherein the VMs include pod VMs, the pod VMs including container engines supporting execution of containers in the pod VMs, wherein the workload is a pod and the workload specification is a pod specification, and wherein the pod specification is received at a master server of the orchestration control plane.

9. The non-transitory computer readable medium of claim 8 , wherein the pod specification includes first resource constraints for a node, and wherein the master server selects the plurality of nodes as candidate nodes based on the first resource constraints in the pod specification.

10. The non-transitory computer readable medium of claim 9 , wherein the virtualization management server converts the first resource constraints into second resource constraints for a host when selecting the node of the plurality of nodes.

11. The non-transitory computer readable medium of claim 10 , wherein the first resource constraints and the second resource constraints include constraints on at least one of processor, memory, network, and storage.

12. The non-transitory computer readable medium of claim 9 , further comprising:
converting, by the orchestration control plane, the pod specification to a VM specification for a pod VM.

13. A virtualized computing system, comprising:
a host cluster, a virtualization management server, and an orchestration control plane;
the host cluster including a virtualization layer directly executing on hardware platforms of hosts, the virtualization layer supporting execution of virtual machines (VMs), the virtualization layer integrated with the orchestration control plane;
the orchestration control plane configured to:
receive a workload specification for a workload;
select, by a first scheduler of the orchestration control plane, a plurality of nodes as candidates for the workload based on the workload specification, each of the plurality of nodes implemented by a host of the hosts;
send, by the first scheduler through a scheduler extender of the orchestration control plane, the plurality of nodes as candidate nodes to a second scheduler of a virtualization management server managing the host cluster;
select, by the second scheduler, a node of the plurality of nodes, the first scheduler being a slave the second scheduler, which is a master, the master receiving the candidates from the slave and selecting the node from the candidates on behalf of the slave;
receive, at the first scheduler through the scheduler extender, the selected node from the second scheduler; and
deploy, in cooperation with the virtualization management server, the workload on a host in the host cluster implementing the selected node, the orchestration control plane configured to deploy the workload through communication with a controller of a hypervisor of the host, the controller configured to manage a VM having an agent executing therein, the contro

Assignees

• Vmware Inc
• VMware LLC

Inventors

• Jain Pranshu
• Wiggers Maarten
• Venkatasubramanian Rajesh
• Gamage Sahan Bamunavita

Classifications

• G06F9/4881

  Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues · CPC title

• G06F2009/4557

  Distribution of virtual machine instances; Migration and load balancing · CPC title

• G06F9/5027

  the resource being a machine, e.g. CPUs, Servers, Terminals · CPC title

• G06F9/45558 Primary

  Hypervisor-specific management and integration aspects · CPC title