the idea xchange  

a revolving guest blog
Welcome to the idea xchange, a new feature of xchange online. This regularly updated section of our site offers opinions and insight from readers like you. We hope you find it informative and interesting. If you’d like to contribute to this revolving guest blog, please contact xchange Editor in Chief Khali Henderson at khenderson@vpico.com.

After starting out as an innovative metro enterprise service, Ethernet has become the hottest enterprise service market (ahead of IP) and the second hottest metro equipment market (behind only metro WDM). But what does it take for Carrier Ethernet service to replace all the TDM private line service demand and finally go over the top?

It’s helpful to remember that “Carrier Ethernet” means different things to different people.

For example, the MEF – which defines Carrier Ethernet as a service  calls out “robustness” as one of the five attributes of Carrier Ethernet, but does not really specify what that level of robustness ought to be. Similarly, the MEF defines Carrier Ethernet service attributes like loss, latency, jitter, and availability, but does not really nail down precisely what a Carrier Ethernet QoS level ought to be.

This isn’t really a bad thing – after all, service providers and equipment vendors alike need latitude in defining their services and concocting their implementations. Therefore, these definitions are broad by design. But this intentional lack of precision does beg the question as to what kind of Carrier Ethernet is needed to really replace TDM.

TDM is the kind of thing people love to hate and hate to love. For years, end users have ordered  to the tune of $10B plus annually  T1s and T3s for private line service because of the SLAs they love: availability, protection switching speed, and connection performance. They have purchased TDM even though they hate the rigidity of the interface, the cost scaling and the difficulty in adding bandwidth.

To satisfy the T1 and T3 private line service demand with Ethernet, two things need to happen: first off, the service level of this kind of “Carrier Ethernet” needs to be equivalent or very close to the service level available from TDM. Secondly, because TDM service is profitable  being delivered over a network and back office infrastructure that is largely paid for  the cost of delivering Ethernet private line in large quantities needs to come down.

So, the approaches to delivering Ethernet private line have got to change in order to meet those SLA and cost conditions. Ethernet over SONET meets the SLA requirement but doesn’t cut the mustard for a network-wide replacement for TDM private line because it does not provide enough aggregation. Switched Ethernet networking is great for aggregation and low cost, but sometimes not high quality enough for many enterprise applications or for the emerging Ethernet mobile backhaul market for that matter.

Connection-oriented Ethernet and its integration into the packet optical network address both the SLA and the cost issues and provide the best of both worlds. Connection-oriented Ethernet is the most stringent form of Carrier Ethernet and delivers the required service quality because it pre-identifies the traffic path and reserves queue resources for individual Ethernet connections end to end through the network This is a real difference from many Carrier Ethernet networks that offer a less stringent QoS because they either operate in a truly connectionless fashion or provide traffic management across large pools of connections but lose visibility to individual connections.

Now the murky world of traffic management can be a tricky thing to decipher (ok, what isn’t) because every box out there does traffic management and has QoS features like policing, shaping, queuing, blah blah blah, and it can be hard to figure out what’s what. The key is the level at which these functions are provided – the chassis? the card? the port? the connection? the CoS class within the connection? Only by providing traffic management at the Ethernet virtual circuit level can TDM-equivalent private line requirements be met.

Marrying connection-oriented Ethernet with the packet optical environment addresses the cost problem of network-wide Ethernet private line. Packet optical networking platforms are the latest generation in optical networking and tightly integrate connection-oriented Ethernet and metro WDM into a single package. These platforms shake up the entire cost profile of the network by providing nearly infinite bandwidth upgradability with WDM while delivering efficient aggregation and port consolidation with connection-oriented Ethernet.

Interestingly, because the packet optical network proves in for access to a connectionless Ethernet switch for multipoint Ethernet services, it further reduces the cost of the service to then bypass the switch entirely for the point-to-point private line Ethernet services, like TDM services are delivered out of the optical network today. The connectionless switch still provides value for multipoint services, but becomes a third wheel for delivering TDM quality private line services.

Carrier Ethernet has come a long way from its roots of a competitive connectionless enterprise service. Connection-oriented Ethernet and packet optical networking are now coming on the scene and are poised to give end users the Ethernet flexibility and the TDM robustness they desire. They also support a network cost structure that will allow service providers to scale the offering and allow Ethernet to go over the top in the network by replacing TDM private line services. Dealing with the details of that TDM-to-Ethernet transition is a topic for another day.

Sam Lisle is market development director at Fujitsu Network Communications. He can be reached at sam.lisle@us.fujitsu.com.