Carrier Ethernet Need To Increase Availability

by Doug Allen
Fri. April 25, 2008

Carrier Ethernet Services is arguably the hottest sector for telcos these days. New switched metro and long-haul Virtual Private Line Services (VPLS) services from Verizon Business and AT&T are the latest indication that next-gen Ethernet services are here, and are poised for rapid growth.

But the industry remains constrained by a number of factors, not the least of which is market awareness. Even with simple service definitions from the Metro Ethernet Forum (MEF) to fall back on, many potential customers are still having a tough time sorting out all the flavors to choose from, and where to deploy them. David Hold, a senior analyst in Business Network Services at Current Analysis, sat down recently with Telecommunications’ Senior Editor Doug Allen to talk about carrier offerings and their applications.

Telecommunications: What are the primary Ethernet services carriers are offering business customers, and how are they targeting them at different types of business (i.e., small-to-medium-business or enterprise)?

Hold: In terms of general topology, carrier Ethernet services can be divided up into three main categories: Connecting a single building or data center to another (point-to-point); connecting a single building to a cloud type service like the Internet or Private IP VPN services (access), and connecting one or more locations to each other (multipoint). Most Ethernet today is metro area point-to-point. Switched multipoint services, formerly known as Transparent LAN, have been almost exclusively deployed in the metro area, although the new VPLS services are both metro and long-haul, and will soon go international.

Traditionally, Ethernet required fiber to the premises, which limited it to high bandwidth applications for medium to large enterprises. A typical example is connecting data centers at gigabit rates. Another would be connecting regional medical center facilities for high bandwidth medical imaging and off-site storage applications. However, with the advent of Ethernet over copper solutions, branch offices are now candidates for Ethernet access up to about 10Mbps.

On the access side, Ethernet is a promising alternative to TDM for local access circuits. Check out Verizon Business’ Ethernet-based Converged Packet Access service, which is designed to combine multiple access types over a single Ethernet connection.

Telecommunications: How do you segment the market? Do you distinguish between types of services—by application, by transport layer (Layer 2/Layer), SONET or WDM used, speed, dedicated vs. shared, long- haul vs. switched intra-metro, etc.?

Hold: All of the above, depending on the situation. Ethernet is most commonly classified by transport type—at the physical/optical layer it is native fiber, SONET/SDH (more expensive), DWDM (high bandwidth but unprotected) and copper (SHDSL or NxT1); and at L2/L3 it used to be transported over ATM and still is in some places. Today, Ethernet is increasingly running over MPLS in the case of Virtual Private Wire Services (point-point pseudowires) or VPLS (multipoint pseudowire mesh). By geography, you can also easily segment services by metro, wide area inter-city and cross-border/international.

Telecommunications: What kinds of important QoS and security parameters are carriers including with these services? Other attractive incentives could include service and/or applications bundles or special pricing incentives.

Hold: This is an area for further development and exploration. There has been little QoS in traditional Ethernet services, although some services use VLAN tags to separate traffic for prioritization; VPLS services can make use of VLAN tags, and can also support QoS the same as in MPLS using DiffServ codepoint (DSCP) or IP Precedence bit marketing.

Telecommunications: Strategically, where are providers trying to go in terms of business models and service deployment/ technology development (maybe inter-city switched Ethernet service)?

Hold: Ethernet is evolving from its roots as mainly a lower cost interface to a metro-optical transport service, to an any-to-any switched data service on a par with frame, ATM or IP/MPLS-VPN. As a Layer 2 VPN, Ethernet still has a ways to go to be fully competitive with IP/MPLS services, mainly in terms of scalability in large networks, but it can be an attractive migration alternative from legacy frame/ATM or TDM services for those end-user organizations that are unwilling to turn control of IP routing over to a carrier.

Telecommunications: When you evaluate carriers’ business models and service deployment plans discussed in question four above, from a customer perspective, what are the likely implications for customers six to twelve months out if carriers follow through on their plans? How will they impact customer’s network usage patterns and buying habits further out?

Hold: It has been interesting to watch AT&T and Verizon Business as they deploy Ethernet Access and VPLS services. The two carriers tend to mirror each other, with one pulling ahead and the other playing catch up. Some of AT&T’s largest customers have been demanding VPLS for several years, and some have defected to alternative carriers over that issue. In the end, the customers are begging their service providers to offer more than just IP-based services, so there is clearly a role for wide area networking services at Layer 2 beyond frame relay and ATM.

Telecommunications: What are business customers’ primary concerns when adopting carrier Ethernet services and how do you see carriers today and in the future addressing them? Where are they getting the most pushback or having the most real-world trouble developing adequate solutions for these concerns?

Hold: I would say that [the first concern] is getting it in the first place, because availability of fiber can be spotty, especially outside of downtown urban cores, and fiber extension can incur sizable costs and lengthy delays.

[The second issue] is the fact that there are so many different types of services that go under the name Ethernet that it can be confusing to know exactly what you are buying if you are not a networking expert. If you think about frame relay, which was the most successful wide area network service since T1, you did not have to worry about distance charges for example, except for the access circuit, and it ran over T1 so you could get it just about anywhere.

[The third issue] is scalability for wide area multipoint networking services such as VPLS. Many experts recommend keeping the numbers of end-points connected to a VPLS in multiples of a few dozen, which poses challenges for 1000 site networks. Beyond that, you may have to segment the network with routers, which is do-able, but in some ways that can defeat the purpose trying to build a pure Layer 2 VPN.

Hierarchical VPLS services could help, but are not yet in commercial deployment. Scalability of VPLS to large networks, in my opinion, is an issue that has not yet been adequately addressed by the carriers. See which major telecom carriers offer gigabit ethernet in your area.