VXLAN on Nexus NX-OS Flood and Learn

What is VXLAN?

In this documentation, we are going to talk about VXLAN, use case scenarios, and deployment of VXLAN with Multicast flood and learn on a Cisco Nexus 9k. Before jumping to the configuration, let’s talk about some theories. What is VXLAN? VXLAN is a virtual extensible local area network. Technically it is a layer two in a layer three overlay tunnel. VXLAN is industry standard, and you can find more specific details from https://tools.ietf.org/html/rfc7348

VXLAN expands Vlan space. As you know, VLANs use 12-bit space, which gives us 4096 values, but VXLAN uses 24 bits space equal to 16,777,216 almost infinite numbers for us. You can ask a question that, I can expand my VLAN over MPLS VPLS or Q-in-Q VLAN tunnels. That is right, but VXLAN is perfectly fitting to the Datacenter environment. If you want to extend your VLAN from one data center to another, VXLAN is a solution. Also, let’s say you have a three-tier architecture Web tier, Application tier, and Database tier. On condition that you follow traditional expansion, your broadcast domain will get bigger; your spanning-tree domain will enlarge. With a VXLAN deployment, you decrease your broadcast domain size. You do not need a spanning tree for loop prevention. Besides, VXLAN uses layer 3 ECMP. Scaling enhancements are Optimization for the control plane, mac learning, Arp tables, Bum replication.

There are some terminologies on a VXLAN.
1) VNI /VNID – Vxlan network identifier – Replaces the VLAN id.
2) VTEP – VXLAN tunnel endPoint. Cisco uses NVE – Network virtualization edge Logical representation of the VTEP. NVE is the tunnel interface Device that performs VXLAN encap/decap, which could be hardware or software.

3) VXLAN Gateway
A device that forwards traffic between VXLANs
It can be both layer two and layer three forwardings.

We can describe the Basic VXLAN workflow:
Receive ARP from a local Host.
Assume a miss occurs, the next step is: Find the remote VTEP. There are three different ways of finding remote VTEP.

Multicast flood and learn
Ingress replication
MP-BGP L2VPN EVPN

In this discussion, we will deploy a Multicast flood and learn.

Caveats

For the existing fabric, you have to configure your MTU size at least 1554 bytes. Here is a small calculation.

Ipv4 header – 20 bytes
UDP header – 8 bytes
VXLAN header – 8 bytes
Original Ethernet header – 18 byte
Original ethernet payload – 1500 bytes

For the VXLAN underlay network, OSPF/isis/Eigrp can be used. I am going to use OSPF. What are we trying to accomplish? The main reason of IGP in VXLAN deployment is building reachability between VTEPs. Our tunnel endpoint is going to be a loopback IP. Loopback IP must be /32 address and OSPF neighborship will be established between those loopback IP addresses. OSPF network type will be a Point-to-Point, and it will simplify adjacency formation.
Let’s move on to the configuration steps. For this demonstration I used eve-ng. CSR1000V is used as a host device. There are some features needed to be enabled on a Nexus 9k. Those are :

feature nxapi
feature ospf
feature pim
feature vn-segment-vlan-based
feature nv overlay
feature bfd

——–Spine 1———–

interface loopback0
ip address 172.24.1.1/32
ip router ospf 1 area 0.0.0.0
ip pim sparse-mode

router ospf 1
bfd
log-adjacency-changes