The second coming of metro Ethernet: early metro Ethernet deployments have ranged from disappointing to disastrous, but carrier interest is still strong, driven by the need for new services and cost savings, and the inability of Sonet/SDH to deliver either
John C. Tanner
This past October, Reliance Infocomm–one of India’s hottest up-and-coming infrastructure players with its own national fiber-optic backbone connecting more than 600 towns–announced plans to deploy what may well be the world’s biggest metro Ethernet network. The network is planned for commercial launch in the first quarter and will comprise over 1,260 building nodes, each of which will support up to eight fiber rings. Each ring will be able to support Ethernet switches installed in the basements of between 112 and 200 mostly residential buildings.
The Reliance project is just one of many indicators of carriers’ growing interest in metro Ethernet as an option for broadband access services, despite its decidedly niche status compared to dominant technologies like DSL and cable modems. Although it has an impressive list of deployments around the globe and a long list of promised benefits–reduced capex and opex, multi-vendor interoperabitity, service differentiation and provisioning flexibility–metro Ethernet services indeed remain the exception to the rule.
More to the point, a number of pioneering greenfield companies that adopted metro Ethernet early have since crashed and burned, many of them in the US, while some survivors are struggling to recoup the costs of their initial network investment. China’s Great Wall Broadband Network (GWBN), for example, was forced in late 2002 to shift its strategy from network deployment to content provision via portal partnerships after the costs of deployment became too much to bear.
Part of the problem with early metro Ethernet deployments, says Andy Miller, Asia marketing director for Juniper Networks, was that the architecture tended to be one big Layer 2 network of Ethernet switches with limited scalability.
“If there’s no congestion, it runs very well,” Miller says. “But once you get congestion, it’s totally random at that point. Remember fast year when all those Internet worms started up? If I’ve got a Layer 2 network, I’ve got no way to control that.”
It’s not all doom and gloom in the metro Ethernet space, however. Asia has been the most aggressive regions deploying optical Ethernet infrastructures, particularly in Korea, Singapore, China and Australia.
PowerComm, the profitable facilities-based telecoms spin-off of Korean power company KEPCO, chose Gigabit Ethernet to link POPs in its metro HFC networks as a cheaper alternative to Sonet/SDH. Metro Ethernet services have popped up in Hong Kong, with one of them, Hong Kong Broadband Network (HKBN), sporting a triple-play offering of broadband, pay-TV and local telephony.
In Europe, carriers such as Telenor and Deutsche Telecom have announced optical Ethernet initiatives. In the US, carriers such as AT&T and SBC began offering new switched, any-to-any (a.k.a. E-LAN) Ethernet services last year on top of existing point-to-point (a.k.a. Ethernet private line) and Internet access services.
Metro Ethernet business is also picking up in Latin American markets such as Brazil, where Cisco Systems alone has managed to secure seven Ethernet contracts in recent months, though actual rollouts thus far have been few.
While features such as SLAs, QoS and interworking remain concerns, analysts say it’s only a matter of time before such things are resolved as second-generation metro Ethernet services and technologies mature into Tier 1 carrier-grade solutions.
Access all areas
One of the more significant shifts in metro Ethernet deployments, says Miller, is a growing tendency by carriers to deploy Ethernet as an access rather than backbone technology.
“A market like Hong Kong is an exception because it’s compact, so it’s possible to create a native environment for Ethernet. For just about anywhere else though, you really want to have a single, manageable metro network to do that, and that’s rare,” he says. “Carriers deploying Ethernet services today are doing it on an IP router core. They start with an MPLS core, and then run Layer 2/Layer 3 traffic across it. It’s an access network and a service that customers want today–they don’t want to build a network with it.”
Frankie Sum, North Asia managing director for Cisco Systems, adds that the Ethernet technology used in the access network depends on how old the in-building wiring is. “In Hong Kong, for example, HKBN uses direct Category 5 connections in the building, with Ethernet running from a hub to the building to the unit, while [CLEC] New World Telecom’s Vitamin BB service uses long-range Ethernet, which uses VDSL to service old buildings wired with older cabling.”
Sum says the distinction is important because “if you’re dealing with older buildings and older wiring, you don’t want to rip that up and replace it if you can help it, so you want to use a technology that works with the old wires.”
Sum adds that it’s not just service providers that are keen on metro Ethernet–private organizations such as banks and universities are also using metro Ethernet gear to build metro WANs for internal use, thanks to the price of equipment coming down over the last couple of years.
“Banks used to rely more on frame relay, which is reliable but expensive,” he explains. “Metro Ethernet solutions make the cost much more attractive.”
Alex Saunders, CEO of New Hampshire-based Metrobility, agrees. “It’s mainly access-based services today. As we get deeper into our customer base, we see that they want more than pure Ethernet. They also want to provide storage services and OC-3 connectivity, and you need Ethernet over Sonet/SDH to do that.”
Which brings up another growing trend–carriers are more keen these days on optical Ethernet solutions that work with (rather than replace) existing Sonet/SDH and/or WDM gear in the metro.
Metro Ethernet strategies
Yankee Group analyst Pat Matthews observes that carriers are moving forward with optical Ethernet infrastructures with two distinct strategies: “building new optical rings or point-to-point infrastructures within the metro edge access points for current demand, and using a cap-and-grow strategy by replacing legacy add/drop multiplexers with newer multiservice provisioning platforms [MSPPs].”
Optical Ethernet solutions come in several flavors, such as Ethernet switching, packet over Sonet/ATM, Ethernet over Sonet/SDH (via protocols like virtual concatenation, generic framing procedure and X.86), Ethernet over RPR (resilient packet ring) and Ethernet over dense or coarse WDM.
Each has its pros and cons (see “How to move your Ethernet on the metro,” below). Ethernet over ATM, for instance, benefits from Sonet/SDH protection but has been criticized for being expensive and having scalability issues. Meanwhile, Ethernet over WDM, regarded by many in the industry as future-proof and strikingly bandwidth-efficient, has problems with reach and overlay, and can be expensive to implement, although Yankee Group rates Ethernet over CWDM as the one optical Ethernet solution that offers the highest overall scalability for the lowest cost.
Matthews says that Ethernet-over-wave sales were on track for 19% year-on-year growth in 2003, compared with growth of just 9% for Ethernet over Sonet/SDH, including RPR. However, he adds, “RPR revenue will increase at the end of 2004 into the beginning of 2005 as it becomes standardized and carriers start to implement this functionality.”
In any case, he adds, with the capex slowdown from carriers likely bottoming out this year and next, there’s plenty of room for growth in the overall market with such a small installed base.
Ethernet over Sonet/SDH is expected to continue to grow, accounting for 22% of MSPP equipment sales revenues by 2007, but the long-term success story will be with Ethernet-over-wave, which will account for well over one-third of WDM product sales by 2007.
“The future looks the brightest for Ethernet-over-wave solutions in the end because it has the widest target audience and within the metro fiber will continue to be a scarce commodity,” Matthews predicts. “Ethernet-over-wave will eventually benefit in 2006 to 2007 from transporting 10-Gigabit Ethernet once the prices for this technology fall.”
Not that it will be that easy. Like any emerging technology, metro Ethernet still has its downsides. Until recently, common complaints from carriers have included higher test and measurement expenses compared to Sonet/SDH due to the need for new test kits and staff to analyze data, dissatisfaction with vendor EMS and OAM&P capabilities, and lack of support for carrier-grade functions such as SLAs and interworking with existing frame relay, ATM and IPVPN services.
However, metro Ethernet is still a work in progress, with standards on the way that address many of these issues. The Metro Ethernet Forum, for example, is working to define six service types for metro Ethernet service providers, such as Ethernet private line, Ethernet relay, Ethernet multipoint services (also known as virtual private LAN service), and Ethernet access to MPLS VPNs (see “In the pipeline,” p. 23). The service definitions are deliberately similar to existing services like private teased lines and frame relay to help carriers offer them as tiered upgrades rather than cannibalize their current business.
Meanwhile, several new standards are in the works that deal with the ability to run Ethernet over copper via VDSL, and point-to-multipoint Ethernet passive optical networks (PONs).
First off the rack this year will be 802.3ah from the IEEE, which adds new standardized features for metro Ethernet such as remote optical link loopback, bandwidth provisioning, the ability to monitor laser transmit and receive power, quality-of-line statistics, and line protection and restoration.
802.3ah also provides a demarcation point between the customer switch or router and the service provider’s network.
“Carriers want end to-end solutions that overcome maintenance, monitoring and provisioning issues at the network demarcation point,” says Matthews. “Embedded remote loop-back, testing and monitoring details for media converters or switches are key to helping carriers with troubleshooting, without costly truck rolls.”
Saunders of Metrobility adds that providing a demarcation point makes IP address management easier. “The demarcation point doesn’t require an IP address, which means there’s no need to manage any extra IP addresses, plus it provides an extra layer of security because there’s no IP address at that demarcation point to exploit,” he says.
How to move your Ethernet on the metro
Technology Pros Cons
Pocket over * Well understood * HDL framing issues with
Sonet (poS) or * Protected by Sonet high-speed
ATM * Currently present in * Cell tax issue
Ethernet over * Standards-based * Based on TDM
Sonet/SDH infrastructure infrastructure
(VCAT,X.86 or * Sonet/SDH resiliency * Requires mapping to TDM
GPS) * Inherent OAM * Point-to-point connections
Ethernet over * Complements Sonet/SDH * Non-standardized
RPR and Ethernet * Provides ring-based
technologies protection only
* Enables efficient
Ethernet over * Future proof * Expensive to implement
WDM * Efficient use of * Less dense of a platform
bandwith * Reach and overlay issues
* Protocol and bit rate
Ethernet * Native delivery of * Lacks protection mechanism
Switching Ethernet * Requires fiber
* Rate adaptive * Lack of OAM capabilities
In the pipeline
Ethernet private line: Like a leased line, only Ethernet.
Ethernet relay: Frame relay with Ethernet interfaces, basically.
Ethernet wire: Provides more efficient bandwith use than Ethernet private line, but without the QoS qualities of Sonet/SDH.
Ethernet multipoint service: Provides any-to-any Ethernet connections. Also known as virtual private LAN service (VPLS).
Ethernet relay multipoint service: an enhanced frame relay service that enables any-to-any topology VPNs and multiple virtual connections consolidated into a single connection.
Ethernet access to MPLS VPNs: A Layer 3 VPN service, where users outsource their router network and IP addressing scheme to the carrier.
10Pass-TS: Uses VDSL to run 10-Mbps symmeterical Ethernet connections over copper loops at distances of at least 750 meters.
2Base-TL: Similar to 10Pass-S, but trades throughput for distance, offering a minimum reach of 2,700 meters but a symmeterical throughput of 2 Mbps. However, it does allow for higher speeds through the use of multiple copper pairs.
100Base-LX10/BX10: For point-to-multipoint Ethernet PONs, with a maximum distance of 10 km on single-mode speeds.
1000 Base-PX10/20: Defines a PON with a splitter capable of serving 16 locations, with distances up to 20 km.
Source: Metro Etherney Forum, Ethernet in the First Mile Alliance (EFMA)
COPYRIGHT 2004 Advanstar Communications, Inc.
COPYRIGHT 2004 Gale Group