EVPN in the Data Center

Chapter 4. Bridging with Ethernet VPN

The first few chapters of this book provided the background and framework for Ethernet VPN (EVPN).

EVPN’s primary function is to support virtual Layer 2 (L2) network overlays. Bridging is how packet forwarding works in L2 networks. Therefore, in this chapter our focus is on how bridging works in EVPN networks. This study includes understanding control-plane and data-plane semantics associated with bridging. The details include a study of choices when it comes to handling frames destined to multiple Network Virtualization Edges (NVEs), handling Media Access Control (MAC) movement, and L2 loops. We study how EVPN works with multiattached endpoints such as dual-attached servers, a common deployment model with EVPN. We conclude with a study of Address Resolution Protocol (ARP) suppression and the benefits it provides.

An Overview of Traditional Bridging

To understand how bridging works with EVPN, we begin with a quick study of how bridging works in a traditional 802.1Q bridge. IEEE 802.1Q defined the standard that describes how VLAN-aware bridges work, and so traditional bridging is also called 802.1Q bridging or plain .1Q bridging. We also introduce the terminology commonly used in bridging. We then use this as the basis to describe how EVPN solves the same problem, albeit differently. A compare-and-contrast table at the end helps solidify these concepts further.

For the sake of this discussion, let us assume the network shown in Figure 4-1 ...