Create a VM with specific CPU and memory resources using VM Runtime on GDC

VM Runtime on GDC lets you configure custom VM types or manually specify CPU and memory resources to match your workload requirements. This document shows you how to create a VM with specific CPU and memory allocations, and how to define and use reusable VM types.

Before you begin

To complete this document, you need access to the following resources:

Access to Google Distributed Cloud version 1.12.0 (anthosBareMetalVersion: 1.12.0) or higher cluster. You can use any cluster type capable of running workloads. If needed, try Google Distributed Cloud on Compute Engine or see the cluster creation overview.
The virtctl client tool installed as a plugin for kubectl. If needed, install the virtctl client tool.

Create a VM

When you create a VM, you can manually specify the CPU and memory requirements. This ability lets you create VMs with the appropriate compute resources to match your application needs.

To create a VM and manually specify the CPU and memory requirements, use the following steps.

CLI

Use kubectl to create a VM:
```
kubectl virt create vm VM_NAME \
  --image ubuntu20.04 \
  --cpu CPU_NUMBER \
  --memory MEMORY_SIZE
```
Replace the following values:
- VM_NAME: the name for your VM. For more information on name constraints, see Object names and IDs.
- CPU_NUMBER: The number of virtual CPUs (vCPUs)to assign to the VM.
  - You can assign between 1 and 96 vCPUs to a VM.
- MEMORY_SIZE: The amount of memory to assign to the VM.
  - You can assign between 1M and 1T of memory to a VM. For more information, see Memory resource units.
Note: In the ~/google-virtctl directory, a VM_NAME.yaml file is created. You can view the content of this file to see the definition of the Kubernetes resources that were created by the VM Runtime on GDC.

Manifest

Create a VirtualMachine manifest, such as my-custom-vm.yaml, in the editor of your choice:
```
nano my-custom-vm.yaml
```
Copy and paste the following YAML manifest:
```
apiVersion: vm.cluster.gke.io/v1
kind: VirtualMachine
metadata:
  name: VM_NAME
spec:
  compute:
    cpu:
      vcpus: VCPU_NUMBER
    memory:
      capacity: MEMORY_SIZE
  interfaces:
    - name: eth0
      networkName: pod-network
      default: true
  disks:
    - virtualMachineDiskName: VM_NAME-boot-dv
      boot: true
```
In this YAML file, define the following settings:
- VM_NAME: the name for your VM. For more information on name constraints, see Object names and IDs.
- VCPU_NUMBER: The number of vCPUs to assign to the VM.
  - You can assign between 1 and 96 vCPUs to a VM.
- MEMORY_SIZE: The amount of memory to assign to the VM.
  - You can assign between 1M and 1T of memory to a VM. For more information, see Memory resource units.
The VM connects eth0 to the default pod-network network.

The boot disk named VM_NAME-boot-dv must already exist. For more information, see Create and manage virtual disks.
Save and close the VM manifest in your editor.
Create the VM using kubectl:
```
kubectl apply -f my-custom-vm.yaml
```

Create and use VM types

When you enable VM Runtime on GDC, a new VirtualMachineType custom resource definition is available. This definition is used to specify CPU and memory resources of a VM. You can create VM types for the different workloads you need, and apply a consistent set of compute resources to VMs based on these types.

If VM Runtime on GDC is enabled in Google Distributed Cloud, the vm-controller-manager installs a predefined VM type. The following definition shows the default example-machinetype VM type:

  apiVersion: vm.cluster.gke.io/v1
  kind: VirtualMachineType
  metadata:
    name: "example-machinetype"
    labels:
      vm.cluster.gke.io/predefined-machinetype: "true"
  spec:
    cpu:
      vcpus: 2
    memory:
      capacity: 4G

You can't update this predefined VM type. This predefined VM type is re-installed if it doesn't exist in the cluster every time the vm-controller-manager is started or restarted, such as if you deleted the VM type.

Create a VM type

You can create your own VM types to fit the compute needs of your workloads.

Create a VirtualMachineType manifest such as my-vm-type.yaml, in the editor of your choice:
```
nano my-vm-type.yaml
```
Copy and paste the following YAML manifest:
```
apiVersion: vm.cluster.gke.io/v1
kind: VirtualMachineType
metadata:
  name: my-vm-type
spec:
  cpu:
    vcpus: VCPU_NUMBER
  memory:
    capacity: MEMORY_SIZE
```
In this VM type, you define the following settings:
- VM_NAME: the name for your VM. For more information on name constraints, see Object names and IDs.
- VCPU_NUMBER: The number of vCPUs to assign to the VM.
  - You can assign between 1 and 96 vCPUs to a VM.
- MEMORY_SIZE: The amount of memory to assign to the VM.
  - You can assign between 1M and 1T of memory to a VM. For more information, see Memory resource units.
Save and close the VM type manifest in your editor.
Create the VM type using kubectl:
```
kubectl apply -f my-vm-type.yaml
```

Create a VM using a VM type

Specify a VM type in your VirtualMachine manifest to apply predefined compute settings to your VM.

Create a VirtualMachine manifest, such as my-custom-vm.yaml, in the editor of your choice.
```
nano my-custom-vm.yaml
```
Copy and paste the following YAML manifest:
```
apiVersion: vm.cluster.gke.io/v1
kind: VirtualMachine
metadata:
  name: VM_NAME
spec:
  compute:
    virtualMachineTypeName: my-vm-type
  interfaces:
    - name: eth0
      networkName: pod-network
      default: true
  disks:
    - virtualMachineDiskName: VM_NAME-boot-dv
      boot: true
```
In this YAML file, specify the name of your custom VM type that you created in the previous section, such as my-vm-type, as the value for the virtualMachineTypeName.

The VM connects eth0 to the default pod-network network.

The boot disk named VM_NAME-boot-dv must already exist. For more information, see Create and manage virtual disks.
Save and close the VM manifest in your editor.
Create the VM using kubectl:
```
kubectl apply -f my-custom-vm.yaml
```

What's next

Edit a VM in Google Distributed Cloud.
When you no longer need VMs, Delete a VM in Google Distributed Cloud.