Starting with VXLAN gateways

Starting with VXLAN gateways

So we talked about the basics on VXLAN now let's talk about the VXLAN gateways and why we required these in our network

VXLAN gateways facilitate communication between a virtual extensible LAN (VXLAN) and another network, particularly a virtual LAN (VLAN).VXLAN traffic must pass through a VXLAN gateway before being able to access remote physical devices.

Connecting VXLAN and VLAN segments with a VXLAN gateway allows tenants to reside in both environments without the need for a separate routing domain.

There are two types of VXLAN Gateways and these are Layer 2 Gateways and VXLAN Layer 3 Gateways.

Fig 1.1- VXLAN Gateways

VXLAN Layer 2 Gateway
VXLAN layer 2 gateways encapsulate Ethernet frames into VXLAN frames and decapsulate VXLAN frames into Ethernet frames. Through the use of a gateway device, a device that does not support VXLAN can be transparently provided VXLAN benefits; this could be a physical host or a virtual machine. 

VXLAN encapsulation is not visible to physical hosts or VMs.

VXLAN Layer 3 Gateways
VXLAN routers are required to communicate between devices in different VXLAN segments, much like traditional routing between VLANs. It transfers frames from one VNI to another.

A frame might need to be decapsulated and re-encapsulated depending on the source and destination. Decapsulation, routing, and encapsulation are supported across the Cisco Nexus device. It is also possible to implement routing across Layer 3 and VXLAN interfaces.

It is possible to enable VXLAN routing on Cisco Nexus device aggregation nodes or at the aggregation layer. During the spine's forwarding, only IP traffic is forwarded, and encapsulated packets are ignored.

A set of VNIs can be grouped into a virtual routing and forwarding (VRF) instance (tenant VRF) to enable routing among those VNIs. If routing must be enabled among a large number of VNIs, you might need to split the VNIs between several VXLAN routers. Each router is responsible for a set of VNIs and a respective subnet. Redundancy is achieved with FHRP.

No comments