VXLAN, or Virtual Extensible LAN, is a network virtualization technology widely used on large Layer 2 networks. VXLAN establishes a logical tunnel between the source and destination network devices, through which it uses MAC-in-UDP encapsulation for packets. Specifically, it encapsulates original Ethernet frames sent by a VM into UDP packets. It then encapsulates the UDP packets with the IP header and Ethernet header of the physical network as outer headers, enabling these packets to be routed across the network like common IP packets. This frees VMs on the Layer 2 network from the structural limitations of the Layer 2 and Layer 3 networks.
Why Do We Need VXLAN?
Why do we need VXLAN? Under the trend of server virtualization, dynamic VM migration occurs, which requires IP addresses and MAC addresses to remain unchanged before and after migration. Server virtualization also leads to a sharp increase in the number of tenants, which the network needs to effectively isolate.
Dynamic VM Migration
Traditional server virtualization works by virtualizing a physical server into multiple logical servers known as VMs. Server virtualization is an effective way of improving server efficiency while reducing energy consumption and operational costs. Such advantages account for its wide use.
Since server virtualization was widely adopted, dynamic VM migration has become increasingly common. To ensure service continuity during the migration of a VM, the VM's IP address and running status (for example, the TCP session status) must remain unchanged. Therefore, VMs can only be dynamically migrated in the same Layer 2 domain.
As shown in the following figure, the traditional three-layer network architecture limits the dynamic VM migration scope. VMs can only migrate within a limited scope, greatly restricting application.
Traditional three-layer network architecture limiting the dynamic VM migration scope
For more information, see VXLAN
