|
Current Issue: September 2007
Feature
Satisfying a taste for Carrier Ethernet
To succeed,Carrier Ethernet’s different flavors need faithful, persistent backers
by Mark Lum
As carriers and suppliers strive to transition from today’s legacy circuit paradigm to tomorrow’s packet future, the noise surrounding Carrier Ethernet gets louder and louder. Its strongest proponents boldly declare the replacement of nearly two decades of SDH/SONET and one decade of IP MPLS network deployment. As far as these backers are concerned, the future is here and it’s called Carrier Ethernet.
The Metro Ethernet Forum (MEF) adopted the term Carrier Ethernet just over two years ago when few, if any, carriers even mentioned Ethernet in their strategic planning for network evolution. Now the industry is buzzing with technology-inspired debate: Vendors offer multiple solutions, while carriers evaluate their best path forward to support next-generation transformation.
Is the Carrier Ethernet market suffering from over-inflated expectations? Many say yes. It may also be the case that so much promotional activity around Carrier Ethernet, from different sources, has simply confused the market.
On the other hand, we might be witnessing the birth of another network disruption just 10 years after the EDFA optical amplifier and WDM technology upset the long-haul transmission network by slashing capacity costs. Only this time, the switching network is next in line for disruption.
Conundrum & paradox
Let’s first consider the long-standing Ethernet conundrum: Is Ethernet a service, a network, or simply an interface? Or is Ethernet all three, even at the same time?
What does Carrier Ethernet actually mean? Is it simply Ethernet used by carriers or has it a more specific meaning? Should we even be concerned with nailing down a definition? Will it give us insight, or can we dismiss the quest for definition as a hair-splitting exercise?
From the MEF’s flag-bearing perspective, things are clear: "Carrier Ethernet is a ubiquitous, standardized, carrier-class service defined by five attributes that distinguish Carrier Ethernet from familiar LAN-based Ethernet." The five attributes are standardized services, scalability, reliability, QoS and service management.
Ethernet services in themselves are hardly new. Most sizeable carriers have provided them since the early 1990s, typically positioned as a cheap, cheerful, fiber-based, short distance LAN extension service. The more serious high value enterprise connectivity services are provided on a mixture of private lines, frame relay or ATM in some cases.
However, the rising tide of sophisticated IP services, increased bandwidth demand and strategic importance of IT has driven a corresponding increase in Carrier Ethernet service expectations. In this area, the MEF leads in creating a clear, common direction for the industry.
Today, a diverse set of networks is used to support Ethernet services, including SDH/SONET, OTN/WDM and VPLS/MPLS, and an equally diverse set of physical media is used to deliver them, including fiber, increasingly copper, and even wireless. Since the turn of the new century, and accelerating over recent years, vendors and operators alike eagerly have embraced such services and service-enabling equipment in their quest to provide new offerings and to support their transformation to an all-packet future.
This perspective shows that Carrier Ethernet also encompasses a broad range of network technologies, with equipment manufacturers enthusiastically adding various Ethernet capabilities to their optical, SDH/SONET and MPLS gear. Given the breadth of definition, perhaps it’s not too surprising that many analysts forecast Carrier Ethernet equipment to grow strongly over the coming years. More to the point, however, it’s clear the ready availability of Ethernet across heterogeneous network platforms is a strong factor in its ongoing success (see Figure 1. Worldwide Ethernet & IP MPLS VPN services revenue forecast, right).
From a purist perspective, however, a Carrier Ethernet network should comprise no more than Ethernet switches acting on the IEEE 802.1x frame information, with no additional headers or labels (such as MPLS) used for switching, and no additional encapsulation (such as GFP) used for transport: a "pure Ethernet" network, if you will, that promises to deliver simplicity, low-cost and ubiquity to its users.
Such networks were indeed deployed by carriers to support private metro and VPN services for enterprises. However, well-known limitations in scalability (the infamous 4094 VLAN limit), manageability (Ethernet having no intrinsic OAM capability), security (broadcast/flooding mechanism) and reliability (spanning tree protocol and proprietary protection schemes) meant that deployments and market momentum stalled several years ago. The situation was not helped by the scarcity of carrier-class equipment practice (see Figure 2. Historical perspective on Carrier Ethernet evolution, right).
While the IEEE 802.x standards groups worked first on PB (provider bridging) and PBB (provider backbone bridging) standards to improve scalability, and then EFM (Ethernet in the first mile) and CFM (connectivity fault management) to improve OAM, VPLS/MPLS networks made rapid inroads in meeting carriers’ requirements for new packet-based infrastructure. New facilities, such as service assurance and resource management, are making VPLS/MPLS systems still more powerful. In parallel, Ethernet-equipped NG-SDH and WDM systems allowed carriers to leverage their existing infrastructures and continue their operational practices and management systems, while still providing Ethernet services to their customers.
So it is that, in addition to the Ethernet conundrum, we have the Ethernet paradox: Ethernet services have never been more popular, but they are not being deployed on Ethernet networks. The mainstream underlying network technology is based on VPLS/MPLS for Ethernet multipoint and virtual services, SDH/SONET for Ethernet private lines and OTN/WDM for wholesale capacity. Pure Ethernet is something of a sideshow, despite there being some significant network deployments that demonstrate the ability and utility of Ethernet for carrier applications without depending on an underlying supporting network.
Packet transport networks transition
From a technological perspective, the most far-reaching changes in network infrastructure are driven by the wholesale transition from circuit-based services to packet-based services. It’s plain that new technology is needed to handle the migration and growth. The continuing increase in packet traffic volume, driven by broadband and other new services, means a packet-based network increasingly becomes economically desirable by many carriers.
The decline of TDM-based networking (e.g., SDH/SONET) has been anticipated over the past decade. While the vast majority of packet traffic still is transported in TDM networks, the inflection point now appears to be in sight as new packet networking techniques come to the fore.
What options are on the table for carriers wanting to evolve their networks and simultaneously deliver new services (e.g., IPTV), cut operational costs and handle increasing amounts of packet traffic? How can carriers make their best choice in packet networking? In reality, such a complex area cannot be black and white. A pragmatic approach is essential if carriers are to have a profitable business.
Choices will be dictated by previous investment, future strategy, targeted services and commercial considerations such as incumbency, costs and second-sourcing. However, four broad packet deployment strategies can be identified (see Figure 3. NGN generic network architecture, right):
• Add packet transport to SDH/SONET. This is essentially "job done" as NGSDH already has received strong support across the vendor community. Most system platforms can be equipped with a wide range of interfaces, aggregation blades, and often L2 switches based on Ethernet, MPLS or even RPR functionality. Networks have been deployed over the past several years.
• Add packet transport to OTN. This is in mainstream development by system houses. A whole range of optical/packet implementations can be identified, from basic Ethernet provisioning on metro or long-haul WDM systems, through integration of L2 switching (based on native Ethernet or even PBT) into WDM, to completely converged OTN/SDH/SONET/MPLS next-generation transport platforms.
• Take VPLS/MPLS to the edge. This is in mainstream deployment by carriers to support services ranging from business Ethernet to residential IPTV. Concerns still exist regarding costs and complexity as the network is scaled up, but new product iterations, commercial competitiveness and support from the packet transport network are expected to alleviate these concerns.
• Take Ethernet/PBB to the core. For carriers that previously invested in a significant Metro Ethernet deployment, this makes sense. New PBB and OAM functionality significantly aids carrier deployment. This approach will appeal to carriers wishing to focus on Ethernet-only services or wanting to explore new options such as PBB-TE or that are particularly concerned about the complexity of IP MPLS.
Into the future: diversity rules
How will the Carrier Ethernet market develop over the next few years? It’s easy to sit on the fence and predict all these technologies will find some place in the carriers’ networks–but that is the most likely outcome. There is sufficient commercial clout behind each different approach to make it a viable solution for carrier customers. And there is sufficient technical differentiation between each approach so that network architects can select between them on the drawing board. The great unknown is operating costs, and that is carrier-dependent to the point where generalization is not useful.
Nevertheless, at some point, carriers must choose. The best, or perhaps the least worst, choice often depends on the starting point–or, as the old adage goes "if you want to go there, I wouldn’t start from here." Few carriers are in the best starting position and are under constant competitive and shareholder pressure.
VPLS/MPLS has proven itself for sophisticated multiservice-oriented applications and is set for still further expansion. Ethernet and MPLS will continue their integration with OTN/WDM for multipurpose transport-oriented applications. PBB/Ethernet, it seems, is struggling to catch up and trailing in the wake of VPLS/MPLS networks. Its best chance for market growth is to focus on new packet transport applications that have been opened up by PBT and to expand from there.
Unlike SDH/SONET, where there were few historical technology choices and a consequent homogeneity in carriers’ network architectures, the different flavors of Carrier Ethernet require faith and determination to succeed in today’s diverse networks.
Mark Lum is an independent, consulting analyst working in the Carrier Ethernet, optical, WDM/OTN and SDH markets. Contact him at www.marklum.net.
|
