Service providers are well familiar with VPLS as a way to leverage their MPLS infrastructure to effectively provide their customers with extending “Ethernet LANs across WANs”; effectively, creating Ethernet domain stretches. Service providers know that VPLS has its short comings; but overall, it has gained a place in their networks to support certain end customer services.

But EVPNs entered the service provider feature set more recently than VPLS and; the EVPN implementations have been maturing to the point that there are some advantages they bring to a network over VPLS implementations. What are those advantages? Are these significant advantages or not ? – these are the bottom-line questions that customers commonly raise.

Let me mention two main advantages of EVPNs over VPLS. There are others, yes… but, these two clearly fall in the “significant” category.

The first difference is that EVPNs use the control plane to communicate MAC+IP address reachability among PE devices while EVPN’s data plane is similar to VPLS data plane. What does this really mean in simple jargon?  In short, it implies that the multiple PE control planes communicate to each other to exchange the MAC+IP bindings and; since this is done at the control-plane level, there is more freedom for network engineers to filter what is imported or exported at each PE – thus, automatic PE-PE flooding, which is part of VPLS,  is consequently, avoided.

The second difference I like to highlight is that EVPNs allow for an active/active multi-homing scenario from a CE-PE perspective. Unlike VPLS which allows for an active/passive multihoming protection. This feature allows EVPN CE’s to actually run a Link-Aggregation protocol across two distinct PE’s which; consequently, permits the implementation of load balance hashing schemes – very nice.

The above two differences should be sufficient to make those service providers with no EVPNs plans yet to raise their eye brows, shake their heads and think: We need to evaluate EVPNs.

Yardiel