Metro Ethernet is not new: It's been around for more than 20 years in various manifestations.
It started out in the mid-1980s as ISDN's central office-based LAN; then came IEEE 802.6 and SMDS. The early 1990s saw the deployment of point-to-point Metro Ethernet, followed by ATM-based multipoint services. The early 2000s saw the launch of several next-generation carriers offering services based on metro-level optics for Gigabit Ethernet with virtual LANs or next-generation SONET. Other approaches have emerged based on Resilient Packet Ring. Recently we've seen some standardization by way of the Metro Ethernet Forum.
So, where is Metro Ethernet now? Certainly, there is a lot of hype about it, but we have seen over the years that the value of a technology is inversely proportional to the amount of trade-press hype.
Metro Ethernet's main applications have been broadband Internet access and virtual private-line service among corporate sites. Using Metro Ethernet is less expensive than using router blades to connect to SONET links. But interconnection agreements among carriers are hard to find, even with the larger providers. Other solutions abound, including MPLS, IP, VPN and ATM.
Penetration also has been relatively slow, for several reasons. Fiber access is still limited to a fraction of all Class A buildings. The performance and service-level agreements are not yet carrier grade, though progress has been made.
Long-haul Ethernet services are not yet generally available. Carrier-to-carrier interfaces have not yet become widely available, which is an issue for users with locations served by different Metro Ethernet providers. A few years ago there were a number of Metro Ethernet providers, such as Infoport Communications and Yipes, but a tough marketplace showed that it is difficult to compete against providers that have a solid infrastructure developed over decades. Many of the start-ups are gone or have altered the business model they follow.
What it takes to make carrier-level Metro Ethernet real is the deployment of high-reliability, operationally sophisticated network elements throughout the infrastructure. Carriers have deployed a couple of million SONET network elements at a cost of US$100 billion. Carriers are not motivated to declare end-of-life for this equipment.
What technology developers do not understand about the carrier's environment is that the (amortized) cost of equipment is usually 5 percent of the total cost of delivering a service; up to 30 percent of the cost is in the operations support side. Hence, for a technology to be successful, it is less important that the equipment is cheaper or that the bandwidth use is more efficient than that it cuts the operations cost by a healthy amount.
I remain reasonably optimistic about Metro Ethernet's value to enterprise users, but what we need is less hype, fewer acronyms, less emphasis on the cost of the network elements and bandwidth savings, and more emphasis on operational capabilities of the network elements, QoS and security. Stay tuned.