Demystifying Optical Ethernet NetworksMonday, February 01 2010The TDM-to-packet network transformation has been underway in transport/ telecommunications networksfor some years now, fueled primarily by two trends: (a) the advent of triple-play (voice, video, data) forenterprise and residential customers and, lately, the explosion in video and mobile data services, and (b) theevolution in both packet- and transport-network equipment. Figure 1. Three key components of Optical Ethernet: service, transport, PHY, together with the technologiesand standards organizations involved in specifying/developing each component.In this regard, there have been rapid advancements to make packet technologies, such as IP and Ethernet,more “circuit-like”, and to make transport technologies and equipment more dynamic and, thus, “packetfriendly.” These developments have led, over the last few years, to the emergence of a mélange of terms —“optical Ethernet”, “metro optical Ethernet”, “packet-optical transport”, “Carrier Ethernet”, “metroEthernet” — which are often used interchangeably, blurring the distinction between them, and leading toconfusion in industry circles. Our objective here is to define the terms optical Ethernet, Carrier Ethernet,and packet-optical transport, explain their relationships, and show how they all fit together in emergingoptical Ethernet networks.
Versatile Packet NetworkingBefore defining the term “optical Ethernet,” it is useful to point out that the term “Ethernet” itself can applyto any one of the three roles of Ethernet technology: as a service, as a transport technology, and as a PHYlayer (Figure 1). Figure 2. Relationships of the different layers: service layer, transport layer, and PHY layer, and their corresponding entities.An Ethernet service is offered to the end-customer (the enterprise or residential customer), runs end-to-end(customer premise-to-customer premise), and is one in which the traffic flow into/out of the system at thecustomer consists of Ethernet frames. An Ethernet service is thus the Ethernet connectivity betweencustomer equipment. A carrier-grade Ethernet service is one that is scalable (to many MAC addresses andend points), offers QoS (traffic management), reliability (protection), and manageability (OAM andmonitoring), and can span long distances (of MAN/WAN scope; typically tens to thousands of kilometers).Ethernet transport refers to the ability to switch/route Ethernet frames (belonging to an Ethernet service)between network nodes, by setting up/using connection-oriented, traffic engineered paths in the networkwith deterministic performance (QoS, delay, jitter, loss, reliability). In other words, Ethernet transportrefers to the setting up of the “pipe” through which the Ethernet frames travel, and to determining itsrouting within the cloud.Ethernet transport makes it possible to realize connection-oriented Ethernet (COE). COE, in essence, refersto the collection of control-plane protocols and data-plane settings that create a connection- orientedcapability for transferring the frames of an Ethernet service. We mention that Ethernet transport could beprovided either by enhancing Ethernet technology (e.g. as is done in Provider Backbone Bridging withTraffic Engineering, PBB-TE, in the IEEE 802.1Qay standard) or by a different technology (e.g. usingMPLS-TP technology being developed jointly by the IETF & the ITU-T). Both of these forms of transportinvolve switching/routing data frames and are, therefore, referred to as Layer 2 (or L2) transport.It is also possible to embed Ethernet frames in a different transport networking layer, such as the oneprovided by the ITU-T’s G.709 OTN (Optical Transport Network) standard. This form of transportinvolves switching/routing traffic at the optical channel data unit (ODU) level and is, therefore, referred toas Layer 1 (or L1) transport.Ethernet PHY refers to the framing and timing of the actual bits of the Ethernet frame, and theirtransmission over a physical medium — copper wire, coaxial cable, or optical fiber — to connect switchesat the physical layer. Some common Ethernet PHYs are the 1 GE (IEEE 802.3z), 10 GE (IEEE 802.1ae),and 100 GE (IEEE 802.3ba) Ethernet PHYs. Note that Ethernet frames can also be embedded in other PHYframing standards, such as those in the ITU-T’s G.709 OTN (Optical Transport Network) standard.
Optical Ethernet NetworkWith this background, we may now define an Optical Ethernet Network as a network spanning aMAN/WAN that offers a carrier-grade Ethernet service, running over a connection-oriented Ethernet(COE) transport infrastructure over an optical PHY (Figure 2). The optical PHY could be provided eitherby the OTN’s optical channel (OCh), or by an Ethernet PHY running over optics, and may be multiplexedonto a given fiber using CWDM/DWDM technology.A key characteristic of optical Ethernet is that its scope is beyond the enterprise LAN, and spans ametropolitan- area or wide-area network.“Carrier Ethernet” vs Optical EthernetThe term “Carrier Ethernet” was formalized by the work of the MEF (Metro Ethernet Forum) in the 2004-2005 time frame, which defines Carrier Ethernet as “a ubiquitous carrier-grade Ethernet service that has thefollowing five attributes: standardized services, scalability, reliability/ protection, hard QoS, and servicemanagement.” The technical work of the MEF (as described in its specifications) together with thetechnical work of associated standards bodies (ITU-T, IEEE, IETF) enable the functionality and attributesof Carrier Ethernet.The services defined by the MEF are in terms of an Ethernet Virtual Connection (EVC), which is definedas an association of two or more User Network Interfaces (UNIs) at the edge of a metro Ethernet network(MEN) cloud (i.e. subscriber sites), where the exchange of Ethernet service frames is limited to the UNI’sin the EVC. The MEF defines three standardized services: E-Line (a point-topoint EVC), E-LAN (amultipoint-to-multipoint EVC), and E-Tree (a point-to-multipoint “rooted” EVC, where the root(s) cancommunicate with any of the leaves, but the leaves must communicate with each other only via the root). Figure 3. Optical Ethernet Network with the service, transport and PHYcomponents in operation.Scalability refers to a service that scales to millions of UNIs (end-points) and MAC addresses, spanningaccess, local, national, and global networks, with the ability to support a wide bandwidth granularity and
versatile QoS options. Reliability refers to the ability to detect and recover from errors/faults withoutimpacting customers, typically with rapid recovery times, as low as 50ms. Hard QoS implies providingend-to-end performance based on rates, frame loss, delay, and delay variation, and the ability to deliverSLAs that guarantee performance that matches the requirements of voice, video, and data traffic overheterogeneous converged networks. Service management implies having carrier-class OAM, and standards-based, vendor-independent implementations to monitor, diagnose, and manage networks offering CarrierEthernet service.Thus, we see that Carrier Ethernet comprises the service component of optical Ethernet networks (Figure 1,Figure 2, and Figure 5).Packet-Optical TransportPacket-optical transport systems (P-OTS or P-OTP) are a new class of networking platforms that combinethe functions and features of SONET/SDH/OTN ADMs or cross-connects, Ethernet switching andaggregation systems, and WDM/ROADM transport systems into a single network element, thus providing“data-aware optical networking.”A P-OTS network element typically will have ITU-T G.709 OTN support, a COE component, and supportfor WDM. These elements also offer transport of a wide range of client signals — Ethernet (dominant),legacy SONET/ SDH, SAN traffic, IP/ATM, video traffic, and can switch at the wavelength level Figure 3.Optical Ethernet Network with the service, transport and PHYcomponents in operation. (WDM), sub-wavelength (or ODU) level, TDM level (SONET/SDH), and packet level (Ethernet, MPLS). A P-OTSnetwork element enables a carrier, especially in the MAN/WAN, to quickly and cost-effectively changeconnectivity and bandwidth in the network, without knowing about the actual services.Key architectural features of P-OTS elements are: • Universal switching architecture/fabric for switching traffic at different layers (OTN, TDM and packet) • Ability to switch, groom, and manage traffic in its native format (i.e. SONET/SDH traffic as TDM traffic, and IP or Ethernet traffic as packet traffic), thus, allowing for the percentage of each traffic type to vary dynamically (all Ethernet to all SONET/SDH and anything in between, for instance) • Software-selectable ports that can switch between switching SONET/ SDH to switching Ethernet, depending on the trafficEven as this definition is gaining industry consensus, according to research firm Heavy Reading, there arethree architectures that are currently deemed to fall under the packet optical transport umbrella, shown inFigure 4.Thus, P-OTS platforms provide the transport and PHY components of optical Ethernet networks (Figure 5).
Optical Ethernet ApplicationsSo which applications/services are optical Ethernet being used for (or envisaged for) today? Figure 4. Packet-Optical Transport Systems (P-OTS): Architectures in use today.As expected, it is the business or residential services with triple-play applications (voice, video, and data tothe desktop), mobile backhaul applications (where the Ethernet PHY is used between the base-station andthe first switching node, and regular optical Ethernet networks are used in the backhaul and backbonenetworks), and utility infrastructure networks (where oil, gas, water, and electric utilities are transformingtheir aged communication systems into “data-aware” systems that allow for automation of functions suchas billing, monitoring, meter reading).Applications such as software-as-a-service, VoIP, VoD, and hosted unified communications are drivingdemand, as are ICT trends such as virtualization, data center outsourcing, data replication, disasterrecovery, remote backup, and IT outsourcing.
How It All Fits Together Figure 5. Optical Ethernet: How it all fits.Thus, we see that in the trio that are the components of optical Ethernet — service, transport and PHY —Carrier Ethernet provides the service component, packet-optical transport gear provides the transport andPHY component, and the various IETF, IEEE, and ITU-T standards provide the specifications for the PHYlayers, as well as connection-oriented Ethernet (Figure 5).As optical Ethernet evolves over the next few years, there will be further reduction in the layers leading to afused Ethernet-WDM packet transport layer with circuit-like capabilities, and to packet- optical systemsoptimized for it. This allows the providers to handle increasing volume of data traffic, while reducing thenumber of network elements by using Ethernet as the common packet technology in access, aggregation,and core networks.This article was written by Vishal Sharma, Ph.D., Principal Consultant & Technologist, Metanoia, Inc.(Mountain View, CA) and Shahram Davari, MASc, Associate Technical Director, Network Switching,Broadcom Corporation (San Jose, CA). For more information, contact Dr. Sharma at email@example.com, Mr. Davari at firstname.lastname@example.org, or visit http://info.hotims.com/28050-201.