PBT is a way of turning off certain key features in standard Ethernet that is currently up for standardization by the IEEE. In conjunction with Provider Backbone Bridging, which is also known as MAC-in-MAC, another proto-standard before the IEEE at the moment, it enables Ethernet to be used as a transport layer in carrier networks. Its reliance on PBB is such that the IEEE study group working on standardization refers to it as PBB-Traffic Engineering, or PBB-TE.

PBB plus PBT unites the former’s scalability, from MAC-in-MAC encapsulation, with the latter’s determinism from a routing perspective, since features such as Broadcast Unknown and Spanning Tree are turned off. As an entirely Ethernet technology, this is all delivered from Layer-2 switches with the necessary firmware changes, avoiding a proliferation of considerably more expensive routers outside the network core.

In bringing the simplicity and economy of Ethernet into the carrier transport domain, PBT won over UK incumbent BT Group Plc, which earlier this month added Nortel to the list of suppliers for its next-generation network, known as the 21st Century Network, a huge fillip for the resurgent Canadian equipment vendor.

To be chosen as the technology for backhauling from 21CN’s multi-service access nodes to its MPLS core, PBT had to beat a L3 proposal (Ethernet over Pseudowire with an MPLS control plane) and an alternative L2 one (VLAN stacking, or Q-in-Q). However, it still has a significant shortcoming because it can currently only be used for point-to-point connections.

This is sufficient for 80% of requirements, according to Mervyn Kelly, Nortel’s director for Carrier Ethernet in EMEA, with the added bonus that its p2p pipes, not an any-to-any cloud, so it’s less complex for fault finding. He said a further advantage is that PBT is not exclusive, so that on the same link we can run a PBT p2p trunk and PBB any-to-any traffic.

For the time being, Nortel is doing just that, running PBB over traditional Ethernet for the remaining 20% of requirements that cannot be met by PBT. However, PBB over traditional Ethernet still has Spanning Tree turned on, and so lacks determinism. Kelly said Nortel identifies a carrier requirement to enable higher oversubscription on any-to-any links, which is predicated on having the ability to know ahead of time that there will be enough bandwidth for such a move. Therefore the vendor is interested in delivering this deterministic feature, together with PBT’s reliability, on any-to-any services.

If Nortel achieves this goal, it will further strengthen its argument in favor of PBT outside the carrier core, particularly as one rival camp with members including Alcatel and Cisco is touting alternatives for enterprise services such as hierarchical VPLS, or HVPLS, which is L2 VPN with a L3 control plane, as a scalable any-to-any technology.