Chapter 6. L2VPN Service Deployment: Configuration and Verification Techniques
In this chapter, we explore L2VPN overlay services established across the SRv6 transport underlay frameworks we’ve already examined. This chapter outlines fundamental approaches for configuration and methods for confirming the integrity of L2VPN service structures. Keep in mind that the content provided here does not delve extensively into L2VPN services. Instead, this chapter serves as a primer on the transition and deployment of well-established L2VPN technologies within SRv6 transport infrastructures.
If there is a requirement for Layer 2 connectivity, a service provider must set up an L2VPN service using technologies such as Virtual Private LAN Service (VPLS), Virtual Private WAN Service (VPWS), or the more recent advancement in L2VPN technology, Ethernet VPN (EVPN). These overlay services are then transported across a unified core transport network. Transport networks capable of handling L2VPN can offer Ethernet LAN (E-LAN), Ethernet Tree (E-Tree), or Ethernet Line (E-Line) services. In this setup, the service provider maps the incoming customer Layer 2 traffic into bridge domains or establishes point-to-point circuits. These bridge domains or point-to-point circuits are then interconnected across the transport network with other customer site Layer 2 bridge domains or point-to-point circuits. In this way, L2VPN services facilitate the extension of subnets from one end to the other, enabling the provision of managed services such as point-to-point, Internet connectivity, intranet, and extranet services to end customers. Figure 6-1 provides a high-level illustration of how L2VPN overlay services are transported over an SRv6 core network, which will be the primary focus of this chapter.”
Figure 6.1 L2VPN Connectivity Across an SRv6 Core Network
L2VPN (EVPN)
The introduction of Multiprotocol Label Switching (MPLS) in RFC 3031, with its highly efficient and flexible data transportation capabilities through the use of MPLS labels, allowed for fast and efficient forwarding decisions without the need for complex IP lookups at each hop. MPLS is protocol agnostic, meaning it can transport packets of various network protocols, such as IP and Ethernet packets. This flexibility is a key part of MPLS’s utility in creating sophisticated and scalable network services, including Layer 2 virtual private networks (L2VPNs).
MPLS L2VPN services provide a transport mechanism for Layer 2 frames between multiple customer sites across an MPLS backbone. These services essentially allow the extension of customer Layer 2 networks across geographically dispersed locations, making it possible to create a VPN that emulates a single LAN segment to the customer. L2VPNs are built using two main architectures: Virtual Private LAN Service (VPLS) and Virtual Private Wire Service (VPWS). VPLS provides a multipoint-to-multipoint service, emulating an Ethernet LAN, while VPWS offers point-to-point connectivity, similar to a traditional leased line. Both VPLS and VPWS are essentially transported across a core network through the use of MPLS pseudowires (PW). A pseudowire is simply an emulated point-to-point connection established across a packet-switched network (PSN) that uses Label Distribution Protocol (LDP) for setting up the pseudowire circuits.
Over the past 15 years, Ethernet VPN (EVPN) has begun to gain traction with service providers and large enterprises, and it is now seen as the logical evolution from the VPLS architecture for Layer 2 provisioning. Limitations that are inherent to VPLS lack of multipathing and multihoming capabilities, lack of multicast optimization and redundancy, among others are resolved through EVPN.
EVPN uses Border Gateway Protocol (BGP) to address these limitations via the control plane, whereas VPLS is inherently a data plane learning and forwarding solution. BGP as its control plane protocol provides more scalability than the flooding and learning mechanism used in VPLS. EVPN can handle a larger number of endpoints without suffering from the same level of complexity and resource consumption that VPLS might encounter due to network expansion. EVPN is able to multicast traffic more efficiently through the use of inclusive multicast Ethernet routes, eliminating the requirement to flood multicast traffic to all endpoints. VPLS, in contrast, typically floods multicast traffic across the entire Layer 2 domain. The multihoming capabilities inherent with EVPN enable single customer edge (CE) routers to connect to multiple provider edge (PE) devices for increased redundancy and load balancing. More granular traffic isolation through unique route targets (RTs) and route distinguishers (RDs) for different services is an additional benefit of EVPN. This is an improvement over VPLS, which typically relies on a single broadcast domain for all connected sites. EVPN uses BGP to advertise MAC addresses, leading to more optimal forwarding paths and faster convergence in the event of network failures. VPLS, on the other hand, relies on traditional MAC learning, which can be slower to converge. EVPN supports both Layer 2 VPN and Layer 3 VPN services, allowing for integrated routing and bridging in the same service instance. This provides greater flexibility in network design compared to VPLS. EVPN is therefore the superior option for Layer 2 VPN services, offering enhanced robustness, scalability, and flexibility over VPLS.
Metro Ethernet Forum (MEF) is an industry consortium that defines standards for carrier Ethernet services. Within the Metro Ethernet Forum 3.0 (MEF 3.0) umbrella standard, several subscriber and operator services standards are defined. One of them, MEF 6.3, is a specification that outlines various Ethernet services and attributes from the perspective of the subscriber. It defines the characteristics and types of Ethernet services that service providers can offer to their subscribers. Figure 6-2 provides a diagrammatic overview of these subscriber services.
Figure 6.2 MEF 6.3: Subscriber Ethernet Service Definitions
Figure 6-3 illustrates how EVPN services map to the MEF 6.3 subscriber service definitions:
E-LAN service (multipoint-to-multipoint connection): EVPN E-LAN interconnects multiple endpoints in a multipoint-to-multipoint fashion that allows for any-to-any connectivity between customer sites, similar to traditional Ethernet LAN services.
E-Tree service (rooted multipoint connection): EVPN E-tree involves designating certain sites as “root” or “hub” sites and others as “leaf” sites, with traffic flowing from leaf to root and from root to leaf but not directly between leaf sites.
E-Line service (point-to-point connection): EVPN VPWS can be used to create a virtual point-to-point Ethernet service. This is typically achieved by setting up an EVPN instance with only two endpoints to provide a dedicated Ethernet connection between two customer sites
Figure 6.3 EVPN Family: Next-Generation Ethernet Service Solutions