Question about when multicasting is not required on BGP EVPN VxLAN

[goyourmin]

hello !

i considering the introduction of various VxLANs.
I'm studying through various documents and YouTube, While watching YouTube, I saw that it is possible to configure it through BGP, EVPN Vxlan, without the need for multicast.

Q1. VxLAN is one of the many technologies that is possible without multicasting?
Q2. Multicast can also use a lot of load equipment, but is it enough to use VxLAN? And is that an advantage?

I'd like to share the YouTube link, but it seems to be copyrighted.

The search title is "VxLAN BGP EVPN | Part 1/5 | Cisco Nexus 9000 Configuration example Control plane learning".

best regard,

[MHM Cisco World]

@hfakoor222 

I study vxlan 6 months ago I write note, but because I now focus in FirePower I keep it later to finish it. 
what point need for vxlan 
1- full understand the BGP with address family, the BGP is responsible of MAC/IP exchange between VXLAN leaf and Spine 
2- MutliCast, for each VLAN there is one MultiCast serve it 
3-VXLAN packet detail 

after these try point you can  easly get the idea of VXLAN 
so Answer for Your Q, you need MultiCast in VXLAN 

what Idea of MultiCast and VXLAN, the leaf1 have VLANx leaf2 have VLANy 
how Spine know that leaf1 have VLANx and leaf2 have VLANy? through the multicast
the leaf send to Spine to join the multiCast and when join the Spine start forward any (unknown or broadcast ) receive from other Leaf for same VLAN to leaf1 or leaf2

[goyourmin]

Hello. Thanks for the quick reply.

I still need to study more. Can you tell us about the topology information you provided?

[MHM Cisco World]

This my note' I read this way' I can not share all it copy write lol.

But explain topolgy 

Vxlan is based on spine and leaf'

Spine connect all leaf and emulate broadcast over l3 layer.

This l3 layer run opsf and bgp.

Bgp add more attributes which can include mac/ip of host. 

This make leaf know the mac address and/or ip of host conenct to other leaf and hence make leaf such as connect to SW (spine emulate this virtual SW)

[M02@rt37 · Accepted solution]

Hello @goyourmin,

Q1) Traditionally, multicast has been used for broadcast and unknown unicast traffic replication in Layer 2 networks. However, VxLAN can leverage different mechanisms, such as BGP EVPN, to handle the control plane and eliminate the dependency on multicast for traffic replication.

Q2) Multicast can indeed introduce additional load on network equipment, especially in large-scale deployments with a high volume of multicast traffic. By using VxLAN with BGP EVPN, you can reduce or eliminate the need for multicast replication within the overlay network. Instead, BGP EVPN allows for efficient and scalable control plane learning, where the information about MAC addresses and reachability is exchanged between the VxLAN endpoints using BGP. This approach can provide advantages such as reduced network complexity, improved scalability, and easier integration with existing network architectures.

I sugget you Toni Pasanen's books about vxlan -- holy bible as I concerned!

[goyourmin]

Thank you so much for your answer.
sir, Does BGP EVPN mean i don't have to use multicast?
I understood that BGP can act as a control plane with EVPN.

Unfortunately, Toni Pasanen's books seem hard to come by in Korea...

Best regard,

[M02@rt37]

Hello @goyourmin,

"Does BGP EVPN mean i don't have to use multicast?" It is not mandatory to use multicast.

"Unfortunately, Toni Pasanen's books seem hard to come by in Korea..." You could buy e-books.

[MHM Cisco World]

We have multi leaf and spine' each one have specific vlan' how spine know to forward traffic to this leaf because it have specific vlan?

Bgp can not detect it' bgp connect all spine and leafs' the multicast is answer

Leaf send joint for multicast ip for specific vlan 

This infrom spine that that leaf need any frame for that vlan 

Then spine send mac/ip for that vlan via bgp.

So it mandatory to use multicast.

Bgp is handle different part of control plane in vxlan.

[MHM Cisco World · Accepted solution]

VXLAN natively operates on a flood and learn mechanism where BU (Broadcast, Unknown Unicast) traffic in a given VXLAN network is sent over the IP core to every VTEP that has membership in that network. There are two ways to send such traffic: (1) Using IP multicast (2) Using Ingress Replication or Head-end Replication.

The main reasons for using BGP EVPN as the overlay control plane are:

• Standards based —The overlay (VXLAN) and the control plane (BGP) are standards based.

• Implement control-plane MAC learning so that VMs/servers for each tenant have a unique identity across the fabric.

So the control plane is divide 
some serve by multicast (broadcast and unknown)
other server by BGP (exchange MAC/IP of host)