how VPNs are implemented can be captured and brought to market quickly, and at minimal cost.

Service topologies supported include point-to-point, point-to-multipoint and any-to-any models. Implementation of these specific topologies gives service providers and enterprises additional flexibility while improving overall resource utilization and network efficiency. For example, an enterprise customer with a headquarters site and many branch offices would benefit from an Ethernet VPN service implemented in a point-to- multipoint or hub-and-spoke fashion. Leveraging a point-to-multipoint service topology, service providers can restrict the spokes to only communicate through the hub, thereby maximizing the efficiencies of their Ethernet infrastructure and optimizing resource utilization and enhancing data security.

Metro Ethernet supported access deployment models are:

•Single enterprise service access via a dedicated link

•Multiple enterprise service access via a pre-standard IEEE 802.1ad Ethernet access link

•Multiple enterprise service access via an Ethernet Services Unit access ring

This flexibility makes it appropriate for both greenfield build-outs as well as demand-based expansions to existing infrastructures. These same capabilities limit the number of truck-rolls required to add users and adjust service levels across the network.

Scalability and efficiency

The primary responsibility of the Ethernet UNI is service demarcation, yet the Nortel Ethernet UNI goes beyond simple demarcation by allowing service providers and enterprises to deliver multiple services and service types per

port. The UNI encapsulates customer data and adds a unique service label so service providers no longer need to worry about overlapping VLAN-IDs, significantly simplifying operations. The Metro Ethernet Routing Switch 8600 maps customer VLAN IDs to IEEE 802.1ah service identifiers. A transparent UNI is defined when all traffic on a physical port is assigned to a single service ID, while a mapped UNI provides multiple services per physical port to one or more customers. This can be accomplished across thousands of service and customer instances as shown in Table 2.

The Metro Ethernet Routing Switch 8600, when performing the role of a Provider Backbone Bridge, aggregating Provider Bridge traffic that is Q-in-Q tagged, can use either the outer (provider) Q-tag or the combination of inner (customer) and outer (provider) Q-tags for service assignment and classification. This capability gives the service provider superior flexibility for interworking with a wide variety of access devices to support the desired service definitions.

The Metro Ethernet Routing Switch 8600 improves bandwidth efficiency by encapsulating enterprise broadcast traffic inside provider unicast packets. Addi- tionally, the 802.1ah Provider Backbone Bridge implementation uses the MAC addresses of the Ethernet UNIs (ingress ports), rather than customer MACs in the switch forwarding tables. This eliminates the “MAC address explosion” issue by greatly reducing the number of MAC addresses that must be learned and maintained by switches in the service provider’s core infrastructure. Keeping the number of MAC addresses to a minimum reduces the aging out and relearning of MAC addresses, thus enhancing end-to-end performance and making network forwarding far more stable.

The Metro Ethernet Routing Switch 8600 delivers high performance, carrier-class Ethernet switching functions for key service provider and large enterprise applications:

•Broadband managed services to apartments, condominiums, office parks, campuses and more

•Carrier-class Ethernet VPNs for seamless

LAN/MAN/WAN connectivity

•Point-of-presence (PoP) edge-to-core traffic aggregation

•Efficient mobile backhaul of 3G/4G traffic

•Latency-sensitive and high availability data center applications

Table 2. Metro Ethernet Routing Switch 8600 scalability

Scalability parameter

Maximum

 

 

Unique service

16,000,000+

identifiers

 

Customer VLANs

4,000

per UNI port

 

Provider Bridge VLANs

4,000

per UNI port

 

EVPN service instances

30,000

per chassis (e.g. E-LINE,

 

E-LAN, E-TREE)

 

 

 

3