01-27-2020 09:04 AM
Hi,
According to the SDA validated designs and best practices you would create an IP Transit between the fabric sites and deploy a C, B, E and WLC function at every site if you have a L3 WAN in between. But you then have to manually configure the WAN peering with VRF lite for every VN and have to use SXP to exchange SGT information.
Is it also supported to stretch a headquarter fabric site over a L3 WAN (e.g. MPLS VPN) with a manual intervention into the underlay routing so that you have end to end VXLAN encapsulation and SGT information and use the WLC at the HQ (head quarter)?
For Example like this: HQ Border (BGP Peering) -> HQ CE (BGP Peering) -> MPLS WAN - > Site CE (BGP Peering) -> Fabric Edge Switch (BGP Peering).
Regards,
G
Solved! Go to Solution.
01-27-2020 09:59 AM
01-27-2020 09:59 AM
02-03-2020 07:24 AM
gtcat,
You an absolutely stretch a fabric between sites if you have the appropriate MTU, aren't running other services that may not work with VXLAN encap and have no firewall blocking the VXLAN UDP port. There are other design considerations (Internet breakout, site survivability, etc) where a stretched fabric may not make sense, so it would be best to review your design with Cisco.
I have a customer running a stretched fabric across Ireland. They have only Fabric Edge nodes in Cork and Gallway, while they have Fabric Border, Control and Edge nodes in Dublin. They use their provide WAN backbone as their fabric underlay and maintain the VXLAN header encap between all sites. They have been deployed in production like this for 18+ months.
Cheers,
Scott Hodgdon
Senior Technical Marketing Engineer
Enterprise Networking Group
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide