Virtualization (KubeVirt) ¶

Kube-DC leverages KubeVirt to provide powerful virtual machine capabilities alongside traditional container workloads. This document covers the virtualization architecture, features, and how VMs are managed within the platform.

Virtualization Architecture ¶

Kube-DC's virtualization layer is built on KubeVirt, which extends Kubernetes to support virtual machine workloads. This architecture enables consistent management of both containers and VMs through the same API and tooling.

graph TD
    K8s[Kubernetes API] --> KV[KubeVirt Controller]
    K8s --> CDI[Containerized Data Importer]

    KV --> VMI[VM Instances]
    CDI --> DV[Data Volumes]

    VMI --> POD[VM Pods]
    DV --> PVC[Persistent Volumes]

    subgraph "VM Management"
        KV
        VMI
        POD
    end

    subgraph "Storage Management"
        CDI
        DV
        PVC
    end

    UI[Kube-DC Dashboard] --> K8s
    CLI[kubectl/virtctl] --> K8s

Core Components ¶

KubeVirt Controller ¶

The KubeVirt controller manages the lifecycle of virtual machines by:

Translating VM specifications into Kubernetes resources
Scheduling VMs on appropriate nodes
Managing VM state (start, stop, pause, resume)
Providing VM migration capabilities
Handling VM monitoring and health checks

Containerized Data Importer (CDI) ¶

CDI handles storage provisioning for VMs by:

Creating and managing Data Volumes
Importing disk images from HTTP/S3 sources
Converting disk formats as needed
Cloning existing volumes

Data Volumes ¶

Data Volumes serve as the storage backbone for VMs, providing:

Storage allocation for VM disks
Integration with Kubernetes storage classes
Automated provisioning and cleanup

VM Management in Kube-DC ¶

VM Creation and Configuration ¶

Kube-DC allows users to create VMs through YAML definitions or the web UI. VM configurations include:

Example VM Definition:

apiVersion: kubevirt.io/v1
kind: VirtualMachine
metadata:
  name: ubuntu-vm
  namespace: demo
spec:
  running: true
  template:
    spec:
      networks:
      - name: vpc_net_0
        multus:
          default: true
          networkName: default/ovn-demo
      domain:
        devices:
          interfaces:
            - name: vpc_net_0
              bridge: {}
          disks:
          - disk: 
              bus: virtio
            name: root-volume
        cpu:
          cores: 2
        memory:
          guest: 4G
      volumes:
      - dataVolume:
          name: ubuntu-base-img
        name: root-volume

Supported Operating Systems ¶

Kube-DC provides templates for a variety of operating systems:

Ubuntu (20.04, 22.04, 24.04)
Debian
CentOS/RHEL
Fedora
Alpine Linux
FreeBSD
openSUSE
Minimal images (cirros)

Network Integration ¶

VMs in Kube-DC are integrated with the same network architecture as containers:

Each VM can connect to VPC networks via Multus CNI
VMs receive IP addresses from the project's CIDR block
Network policies apply to VMs just like containers
VMs can use floating IPs and load balancer services

Storage Management ¶

Kube-DC provides flexible storage options for VMs:

Support for multiple storage classes
Persistent storage using Kubernetes PVCs
Live volume resizing
Volume snapshots and cloning

VM Customization ¶

VMs can be customized through cloud-init configurations:

cloudInitNoCloud:
  userData: |-
    #cloud-config
    chpasswd: { expire: False }
    password: securepassword
    ssh_pwauth: True
    package_update: true
    package_upgrade: true
    packages:
    - qemu-guest-agent
    runcmd:
    - [ systemctl, start, qemu-guest-agent ]

This allows for: - Setting initial passwords - SSH key distribution - Software installation - Custom scripts execution - Network configuration

Health Monitoring ¶

VMs in Kube-DC support health checks through:

readinessProbe:
  guestAgentPing: {}
  failureThreshold: 10
  initialDelaySeconds: 20
  periodSeconds: 10

Health checks ensure: - VM is properly booted - Guest agent is responsive - Cloud-init has completed - Custom health check scripts pass

Web UI Management ¶

Kube-DC provides an intuitive web interface for VM management:

VM Management UI

VM Dashboard Features ¶

The VM dashboard provides:

VM Status Monitoring: Running status, uptime, and conditions
Performance Metrics: Real-time CPU, memory, and storage usage
VM Details: OS version, network configuration, and node placement
Console Access: Direct web-based console access to VMs
SSH Terminal: Direct SSH access from the browser
Network Information: IP addresses and VPC subnet details

VM Lifecycle Management ¶

Through the UI, administrators and users can:

Create VMs from templates or custom images
Start, stop, pause, and restart VMs
Adjust resource allocations (CPU, memory)
Take snapshots for backup purposes
Clone VMs to create new instances
Migrate VMs between nodes

Advanced Features ¶

GPU Passthrough ¶

Kube-DC supports GPU passthrough for high-performance computing and AI workloads:

domain:
  devices:
    gpus:
    - deviceName: nvidia.com/GP102GL_Tesla_P40
      name: gpu1

Live Migration ¶

VMs can be migrated between nodes without downtime:

spec:
  strategy:
    type: LiveMigrate

VM Snapshots ¶

Kube-DC supports VM snapshots for point-in-time recovery:

apiVersion: snapshot.kubevirt.io/v1alpha1
kind: VirtualMachineSnapshot
metadata:
  name: my-vm-snapshot
spec:
  source:
    apiGroup: kubevirt.io
    kind: VirtualMachine
    name: my-vm

VM Templates ¶

Organization administrators can create standardized VM templates for their users, ensuring consistent deployments and reducing configuration errors.

Integration with Multi-Tenancy ¶

VMs in Kube-DC operate within the same multi-tenant architecture as containers:

VMs are created within specific projects
Organization and project permissions control VM access
Network isolation is enforced between projects
VM metrics are included in project billing and quotas

Best Practices ¶

Resource Allocation ¶

Allocate sufficient memory for the guest OS (minimum 1GB for most Linux distributions)
Consider CPU overcommit ratios when planning node capacity
Use appropriate storage classes for VM performance requirements

VM Optimization ¶

Install guest agents for improved integration
Use cloud-init for automated VM configuration
Configure readiness probes for proper health monitoring
Use virtio drivers for improved performance

Conclusion ¶

Kube-DC's integration of KubeVirt provides a seamless experience for managing both VMs and containers in a single platform. This unified approach simplifies infrastructure management, improves resource utilization, and enables hybrid application architectures that combine the benefits of both virtualization and containerization.