Carrier Ethernet is a hot topic even in an industry that seems to find heat all too easily in some situations.
Ethernet is a mature standard. It's so cheap that consumer and small business switches are sometimes given away as premiums. The standards evolution has pushed its speed from the original 10 Mbps to 10 Gbps, with 100 Gbps promised.
It seems a carrier's dream, but there's a problem. Carrier Ethernet today suffers from a problem of focus, and planners will have to fix that problem to be successful with it.
There have been two avenues by which service providers have traditionally approached Ethernet deployment:
- The "SONET replacement" approach that focused on the use of new-generation packet technology to replace traditional SONET in inter-office and outside plant applications. One of the major areas of interest here was Resilient Packet Ring (RPR).
- The "enterprise Ethernet" approach that focused on the use of Ethernet as an alternative to T3 or SONET in creating access connections for enterprises, usually for headquarters or major data center locations.
Consumer broadband changes the equation
While there is certainly value in both these approaches, the recent interest in carrier Ethernet is spurred by a third factor, which is the explosive growth of consumer broadband access to the Internet and collateral delivery of IPTV. An IPTV application can create 10Gbps or more of incremental traffic to a typical residential central office, far more than most metro enterprise central offices have today. The problem of delivering this kind of bandwidth creates a mission for metro infrastructure unlike anything previously considered. At the same time, it may also drive those earlier Ethernet missions forward at a faster pace.
Residential bandwidth requirements create a scale never before seen in metro infrastructure, and thus an economy of scale never seen as well. That creates a strong incentive to migrate additional traffic from other sources to the same infrastructure used for residential delivery. That may well be carrier Ethernet.
There are two dimensions to the carrier Ethernet metro infrastructure value proposition:
- Ethernet technology is generally far less expensive than IP technology. Network operators put the capital savings at between 30% and 45%, and where residential services are involved, any cost savings can be critical in securing optimum penetration, reducing churn and maximizing profits.
- Ethernet technology is well-suited to the traditional metro mission, which is the creation of star topologies to link COs with either metro off-ramps onto wide-area (and Internet core) services, or metro data centers and content farms. By creating traffic tunnels, consumer and business traffic are separated from each other, and at the same time separated from the control plane and address space of the metro infrastructure itself.
Managing service quality
The infrastructure mission of carrier Ethernet is logical, and its justification is strong, but it is still a departure from the mindset under which most carrier Ethernet products and standards have developed. One indication of this is the recent furor over "Provider Backbone Transport" (PBT), an enhancement to carrier Ethernet to provide for traffic-engineered point-to-point tunnels. PBT, now proceeding through the IEEE as "Provider Backbone Bridging with Traffic Engineering" (PBB-TE), is essential in creating manageable service quality in an Ethernet metro area, but is not yet finalized.
Perhaps the most significant issue for planners in carrier Ethernet is a derivative issue of PBT, which is the notion of a separate control plane. Ethernet bridging was clearly not suitable for metro infrastructure missions, and PBT replaces it with a flexible and separate control plane that can eliminate adaptive route changes in favor of provisioned routes, something operators think critical when melding optical-layer service recovery with that provided by the electrical layers of the network. However, this separate control plane implies a strong management system framework above it to help operators conceptualize services and classes of service (enterprise versus consumer, video versus Internet) and create those classes in a non-competing way on their networks. While carrier Ethernet standards support this kind of management framework, they do not mandate it and some operators are finding it difficult to secure the kind of tools they need, so care must be taken here.
Another issue is Ethernet OAM&P. Management telemetry is being added to carrier Ethernet through efforts of a number of standards bodies, notably the Metro Ethernet Forum (MEF), providing some of the same end-to-end capabilities that SONET offers in its segment and path header structure, for example. However, these OAM&P features are not yet fully integrated with Ethernet enterprise services, and not at all integrated with consumer services delivered through tunnels (L2TP, PPPoE, etc.) over Ethernet infrastructure. Again, this demands management system coordination of multi-layer service management functions, and not all management systems can provide this.
The question of "tunnels" may be of great importance to carrier Ethernet planners. Ethernet, like most Level 2 protocols, does not create a formal network-to-network interface and does not completely isolate the user-network interface control plane from the network control plane. Decisions made at the service level can thus impact the infrastructure, and vice versa. Where carrier Ethernet is used simply to host tunnels, this is not likely to be a problem. But where Ethernet VLAN services are sold to enterprises and the same infrastructure is used to host consumer broadband, the setup for both service sets may be inter-reactive, and careful planning is needed to insure stable operation. Some network operators are looking seriously at deploying tunnels or pseudowires over Ethernet for enterprise services to avoid issues here, but this works best for point-to-point services since there are no standard "multipoint tunnels" yet defined.
One clear lesson from operators is that it is not prudent to allow the processes of enterprise Ethernet service planning, SONET replacement, and consumer broadband metro infrastructure planning to proceed independently unless the three distinct applications are to be kept separate at the Ethernet layer and integrated only at the optical layer. This latter practice may be a reasonable short-term measure to consider while standards and experience both catch up with the explosive interest in Carrier Ethernet as metro infrastructure.
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.