Virtualization and VRFs

A hypervisor runs multiple virtual machines on one physical host. Type 1 (bare-metal, e.g. ESXi, Hyper-V, KVM) runs directly on hardware; Type 2 (e.g. VirtualBox, Workstation) runs as an application on a host OS. Each VM gets virtual CPU, memory, NICs and storage isolated from the others.

A VRF (Virtual Routing and Forwarding) instance is a separate routing table and set of interfaces inside one physical device. Each VRF has its own routes, ARP table and forwarding decisions — interfaces assigned to different VRFs cannot reach each other unless routes are deliberately leaked. One router can therefore host a management VRF, a guest VRF and a production VRF that never mix.

Virtualization also extends to the network itself: hypervisor vSwitches connect VMs to physical NICs, and overlay networks let VMs on different hosts share a virtual segment. On Cisco routers the simplest exam-relevant virtualization is the VRF: ip vrf creates the instance, ip vrf forwarding pins an interface to it, and show ip vrf lists them.

Common mistakes

• Treating a VRF like a VLAN — a VRF separates L3 tables; VLANs separate L2 domains. They solve different problems.
• Pinging without the VRF qualifier after assigning an interface — the global table no longer knows that subnet.
• Assuming containers give VM-grade kernel isolation — they share the host kernel by design.
• Configuring ip vrf forwarding after the IP address — IOS removes the address when the interface joins a VRF; re-apply it.