This document describes how to achieve a Virtual Extensible LAN (VXLAN) Ethernet Virtual Private Network (EVPN) Multi-Site design by integrating VXLAN EVPN fabrics with EVPN Multi-Site architecture for seamless Layer 2 and Layer 3 extension. In addition to the technical details, this document presents design considerations and sample configurations to illustrate the EVPN Multi-Site approach. The VXLAN Border Gateway Protocol (BGP) EVPN fabric (or site) can be extended at Layer 2 and Layer 3 with various technologies. However, the sole focus of this document is on how this extension can be achieved by using EVPN Multi-Site architecture, an integrated interconnectivity approach for VXLAN BGP EVPN fabrics.
EVPN Multi-Site technology is based on IETF draft-sharma-multi-site-evpn.
VXLAN EVPN Multi-Site architecture is independent of the transport network between sites. Nevertheless, this document provides best practices and recommendations for a successful deployment.
where we have deployed it in mega scaling option, which is EVPN-PBB. Its scalability is caused by hiding of all customer’s MAC addresses behind MAC address of PE router for control plane, like in classical PBB deployment. But you still have data plane MAC learning for customer MAC addresses on respective PE. On the other hand, complexity with such deployment is higher than traditional EVPN deployment, where customer MAC address learning happens in control plane. We haven’t reviewed the ordinary EVPN (as well as VXLAN), because Cisco IOS XRv, at least my version doesn’t support it. So in current article I’ll use Nokia (Alcatel-Lucent) VSR as VTEP (virtual termination end point), which is actually leaf switchers terminating VXLAN encapsulation. Cisco IOS XRv routers will play role of data center spine switches. Though the names spine/leaf might be new to you, if you don’t have data center background, their definition can be translated to IP/MPLS world
