Carriers building IP-based next-generation network infrastructures are faced with the tough task of guaranteeing service levels and uptime requirements in a huge, global, but less-than-deterministic network environment.
Other stories on this topic Carriers skimping on enterprise services?
To address this shortcoming of IP, carriers are examining two primary technologies: Transport Multi-protocol Label Switching (T-MPLS) and Nortel's Provider Backbone Transport (PBT). PBT also is called Provider Backbone Bridging-Traffic Engineering (PBB-TE) within the IEEE.
PBT is an Ethernet derivative intended to bring connection-oriented characteristics and deterministic behavior to Ethernet. It turns off Ethernet's Spanning Tree and media-access-control address-flooding and learning characteristics. That lets Ethernet behave more like a traditional carrier transport technology.
The key selling factor in its favor compared to T-MPLS, the traffic-oriented version of MPLS, is that it's based on Ethernet and can be implemented via switches, obviating the need for more costly MPLS-aware routers.
Unfortunately for PBT, it's relatively new and untried, whereas MPLS has been around for 10 years and has the backing of the major router vendors, including network king Cisco.
"If you look at the difference between PBT and T-MPLS, you could make an argument that they provide almost identical features and capabilities," says Thomas Nolle, president of the technology assessment firm CIMI and a Network World columnist.
"But PBT equipment is 33% cheaper. So the carriers are saying, 'Wait a minute. If I can get the same thing for less money, then why do I need the IP routing part?'
"And that's where the router guys are increasingly on the defensive."