Turn One Linux Server into Ten: A Complete KVM/QEMU Virtualization Guide

What is virtualization

Virtualization tricks the operating system into believing it has exclusive access to a full computer while the hypervisor shares the underlying CPU, memory, storage, and network among multiple guest systems.

Core benefits

Higher resource utilization : CPU and memory usage can rise from ~10% to 60‑80%.

Strong isolation : a crash in one VM does not affect others.

Snapshots & migration : instant state capture and near‑zero‑downtime live migration.

Environment consistency : identical images for development and production.

Rapid scaling : new VMs can be cloned in minutes.

Hardware vs software virtualization

Hardware‑assisted virtualization (HVM) uses Intel VT‑x or AMD‑V. Overhead is typically 2‑5% and behaves like bare metal. grep -E 'vmx|svm' /proc/cpuinfo If the output contains vmx (Intel) or svm (AMD) the CPU supports it; otherwise enable the feature in the BIOS.

Software‑only virtualization relies on QEMU to emulate the entire hardware stack, offering universal compatibility but incurring 30‑50% performance loss.

In practice the dominant approach combines both: KVM handles hardware‑assisted scheduling, while QEMU provides device emulation.

Popular virtualization solutions

KVM : kernel‑based hypervisor, near‑bare‑metal performance, managed with libvirt.

VirtualBox : desktop‑focused, GUI‑driven, less suited for production.

VMware ESXi : Type‑1 bare‑metal hypervisor, feature‑rich but costly.

Proxmox VE : open‑source KVM management platform with LXC support.

Docker containers : process‑level isolation sharing the host kernel; not a traditional VM.

KVM environment setup

1. Install required packages

Install qemu-kvm, libvirt, virt-install, and bridge-utils (package names vary by distro). Verify the libvirtd service: systemctl status libvirtd If it is not active, start it:

systemctl start libvirtd

2. Create the first VM

Use virt-install for a concise one‑line definition:

virt-install \
  --name web-server \
  --vcpus 2 \
  --memory 2048 \
  --disk path=/var/lib/libvirt/images/web-server.qcow2,size=20,format=qcow2 \
  --network bridge=br0 \
  --os-variant ubuntu20.04 \
  --graphics vnc \
  --cdrom /path/to/ubuntu-20.04.iso

Key parameters: --name: identifier used by all virsh commands. --vcpus: number of virtual CPUs (do not exceed host threads). --memory: RAM in MB. --disk: path, size, and format (qcow2 supports snapshots and thin provisioning). --network bridge=br0: attach the VM to a bridged network. --os-variant: hints for optimal defaults. --graphics vnc: enable VNC console for installation.

3. Daily VM management (virsh)

virsh list

– list running VMs. virsh list --all – list all VMs. virsh start <name> – power on. virsh shutdown <name> – graceful shutdown. virsh destroy <name> – force power off. virsh reboot <name> – reboot. virsh suspend <name> – pause (memory state kept). virsh resume <name> – resume. virsh autostart <name> – enable start on host boot. virsh console <name> – connect to serial console. virsh dumpxml <name> – export XML definition. virsh undefine <name> – delete configuration (disk remains).

4. Snapshot management

Snapshots require the disk to be in qcow2 format. Example workflow:

# Create snapshot
virsh snapshot-create-as web-server pre-upgrade-snap

# List snapshots
virsh snapshot-list web-server

# Revert to snapshot
virsh snapshot-revert web-server pre-upgrade-snap

# Delete snapshot
virsh snapshot-delete web-server pre-upgrade-snap

5. Cloning VMs

Clone a VM quickly with virt-clone:

# Simple clone with automatic MAC and hostname handling
virt-clone --original web-server --name web-server-2 --auto-clone

# Clone with explicit new disk path
virt-clone --original web-server --name web-server-2 \
  --file /var/lib/libvirt/images/web-server-2.qcow2

Virtual network configuration

Mode 1 – NAT (default)

The default default network provides outbound connectivity but blocks inbound connections.

# Show NAT network info
virsh net-info default

# List DHCP leases (VM IPs)
virsh net-dhcp-leases default

Mode 2 – Bridged (production)

Install bridge utilities:

yum install -y bridge-utils   # CentOS/RHEL
apt install -y bridge-utils   # Ubuntu/Debian

Create bridge br0 and attach the physical NIC (e.g., eth0), then move the IP configuration:

# Create bridge
brctl addbr br0
brctl addif br0 eth0

# Move IP configuration
ip addr del 192.168.1.100/24 dev eth0
ip addr add 192.168.1.100/24 dev br0
ip link set br0 up

Persist configuration (CentOS/RHEL example): create /etc/sysconfig/network-scripts/ifcfg-br0:

TYPE=Bridge
DEVICE=br0
BOOTPROTO=static
IPADDR=192.168.1.100
NETMASK=255.255.255.0
GATEWAY=192.168.1.1
ONBOOT=yes

Add BRIDGE=br0 to the original ifcfg-eth0 and remove its IP lines.

When creating a VM, specify the bridge:

virt-install ... --network bridge=br0 ...

Mode 3 – Host‑Only

Create an isolated internal network reachable only from the host:

virsh net-define /dev/stdin <<EOF
<network>
  <name>hostonly</name>
  <ip address="10.0.0.1" netmask="255.255.255.0">
    <dhcp>
      <range start="10.0.0.100" end="10.0.0.200"/>
    </dhcp>
  </ip>
</network>
EOF
virsh net-start hostonly
virsh net-autostart hostonly

Production pitfalls

Use qcow2, not raw : qcow2 supports snapshots and dynamic expansion; raw lacks these features.

Reserve host memory : keep at least 2 GB free; monitor with virsh dommemstat <name>.

Bridge configuration : move the IP address to the bridge interface; configuring the IP on the physical NIC breaks connectivity.

Proper VM removal : shutdown → virsh undefine <name> → manually delete disks, or use virsh undefine --remove-all-storage (libvirt 4.10+).

Clean up snapshots : snapshots are not backups; excess snapshots degrade disk performance.