One of the most-publicized debates in the telecom space today is the one swirling around the concept of provider backbone enhancements to Ethernet, the set of standards that really establish "Carrier Ethernet" as a reality. It's been a love-hate thing, with router vendors on the hate side and practically everyone else on the love side. This already complex debate is getting even more complicated, which isn't making network planners' lives any easier.
Provider Backbone Bridging (PBB) is an enhancement to standard Ethernet that makes it scalable to very large networks. PBB made Ethernet a practical option as the foundation for metro services everywhere, and even for national networks in some EU geographies, for example. But while PBB might offer enough address scalability to match IP in these deployments, it lacked traffic engineering capability. This was added with the emerging PBB-TE standard, which is more popularly called Provider Backbone Transport (PBT). PBT not only provided Carrier Ethernet with traffic management tunnels, but also with a highly controllable forwarding process that let providers define backup paths and control routes precisely.
PBT's features were touted by some major providers worldwide, including both BT and Deutsche Telekom in Europe. Operator surveys conducted by CIMI Corporation now show that PBT is being actively explored by all 10 of the major worldwide Tier 1 operators regularly tracked by the company. PBT is also the Carrier Ethernet development that got router vendors' backs up, since it competes with MPLS, a concept supported with religious zeal in the IP community.
All communications services popularly used are based on one of three connection models, characterized by the Metro Ethernet Forum as:
- E-Line: A service type used to create Ethernet private lines, virtual private lines and Ethernet Internet access
- E-LAN: A service type used to create multipoint Layer 2 VPNs, transparent LAN service, and the foundation for multicast networks (IPTV and TLS)
- E-Tree: Point-to-multipoint services requiring less configuration than traditional hub-and-spoke configurations using E-Lines
To make a "service," a connection model is combined with a delivery protocol. Ethernet is one such protocol and IP another. But since Ethernet is a Level 2 protocol, it can (and most often does) transport IP. That means you can deliver what the user believed to be an IP service by simply dumping properly formatted packets into the correct connection model for the service you were supporting, with proper formatting, addressing and encapsulation. You could use E-Tree to deliver IPTV, for example.
Needless to say, this is going to cause more angst in the IP world, because operators believe that the use of Carrier Ethernet could save them 30-40% in capital equipment costs. Some are now saying privately that it could also save them considerable operational cost and reduce their personnel training budgets, too (IP networks seem to spawn training and certification programs like weeds).
PBT/Carrier Ethernet rise as Internet changes
Ironically, one thing that's driving PBT and Carrier Ethernet into confrontation with IP and MPLS is the evolution of the Internet. A communications network is designed to support any-to-any connectivity; you can dial any phone number from any phone. The Internet is transitioning into a network that delivers experiences, and those experiences are hosted by a small number of sites. Thus, the Internet as we actually use it, is less and less an "open connectivity" network. In fact, a lot of that open connectivity, when exercised, creates security risks like denial of service (DoS) attacks. Building something like a VPN over Carrier Ethernet would partition that VPN absolutely from the Internet, eliminating any Internet-based threat or risk. It also significantly means that most traffic flows on a few relatively contained routes to these experience hosts, which is ideal for Ethernet.
A second problem Carrier Ethernet supporters say they can solve is the lack of stringent service-level agreements (SLAs) for IP services. IP networks are adaptive in their routing, and during periods of network reconfiguration, services can be interrupted while the network re-learns its topology. BT, for example, cited this adaptive behavior as a reason why converging older Frame Rrelay or ATM services onto IP often meant offering a less stringent SLA because traffic patterns and policies couldn't be managed as tightly. PBT doesn't do anything adaptively. The forwarding tables are completely under control of a higher layer that can be as adaptive or non-adaptive as the network operators wish.
There will always have to be some routers in any IP network, but if Carrier Ethernet manages to tap off significant metro investment, it will have a significant impact on router sales. When valuable content is consumed in a given metro area to justify hosting a copy there, content raises metro bandwidth requirements significantly but has a much smaller impact on core bandwidth. That makes the metro infrastructure market the hottest space in the industry, the one most likely to get big-budget support in the next few years. For anyone building a metro network or supporting a national network that includes significant enterprise traffic with stringent SLA requirements, Carrier Ethernet and PBT are looking more and more like the optimum choice.
About the Author: Tom Nolle is president of CIMI Corporation, a strategic consulting firm specializing in telecommunications and data communications since 1982. He is a member of the IEEE, ACM, Telemanagement Forum, and the IPsphere Forum, and the publisher of Netwatcher, a journal in advanced telecommunications strategy issues. Tom is actively involved in LAN, MAN and WAN issues for both enterprises and service providers and also provides technical consultation to equipment vendors on standards, markets and emerging technologies. Check out his SearchTelecom networking blog Uncommon Wisdom.
