PACKET-OPTICAL THE INFINERA WAY
INFINERA 1 OPTICAL FIBER PROVIDES almost lossless telephone service) to cover a wide range of We hope you find Packet-Optical the PREFACE transmission of signals at an ultra-wide solutions and networks with varying degrees Infinera Way informative and useful, whether range of frequencies. Packet switching, of capabilities and functionality. you use it to research a particular subject or implemented using the Ethernet family read the complete volume from beginning Packet-optical integration has some great of protocols and interfaces, offers one of to end. advantages in terms of cost and service the most efficient ways to sort and direct differentiation. Infinera´s technologies take The descriptions are kept independent streams of digital data. Packet-optical this one step further, with benefits including of product releases as much as possible. networking combines these two outstanding reduced equipment, lower operational costs Current details of the Infinera product technologies, positioning them to dominate and key capabilities such as low latency portfolio are available at www.infinera.com. the next generation of transport networks. and excellent synchronization, outlined Packet-Optical the Infinera Way was written in Chapter 2. Chapter 3 describes how Features of Infinera’s packet-optical solutions to help Infinera’s customers, prospects packet-optical networks are best managed that we believe to be unique are highlighted and partners, and anyone else who needs and how to take advantage of current and with this marker throughout the text. to have a better understanding of the future software-defined networking (SDN) packet-optical world. This book focuses on developments. Chapter 4 takes the reader The information included is subject to change the integration of higher Open Systems further into how the described technologies without further notice. All statements, information and recommendations are believed to be Interconnection (OSI) layer functionality into are leveraged by various applications, such accurate but are presented without warranty of optical systems to create packet-optical as business Ethernet, mobile fronthaul/ any kind. transport systems (P-OTS), i.e. packet- backhaul and cable TV backhaul. optical networks. The term P-OTS is widely For those looking for a better understanding used within the industry and has been of Layer 2 Ethernet technologies, Chapter 5 recycled from an earlier definition (plain old includes a more detailed description of how these technologies work and how they are leveraged in wide area networks.
2 INFINERA Footnote citations on page 111 INFINERA 3 CONTENTS
1 INTRODUCTION...... 6 2.3 MPLS-TP for Traffic Engineering 2.7.1 Benefits of the Packet-Optical 3.3.3 Applications of Intelligent Packet 4.4.2 A Backhaul Network Optimized 5.5.1 Carrier Ethernet: Ethernet as a and Service Scalability...... 26 Approach...... 38 Transport with SDN...... 57 for 4G/LTE and 5G...... 68 Transport Service...... 88 1.1 The Cloud-driven Transformation 2.4 Traffic Management...... 30 2.7.2 Advantages when Using Infinera’s 4.5 Switched Video Transport...... 69 5.5.2 The Carrier Ethernet of the Network...... 8 4 APPLICATIONS OF PACKET- 2.4.1 Traffic Shaping and Policing...... 30 Portfolio for Packet-Optical 4.5.1 Streaming 3D and HD Video to Architecture and Terminology...... 91 1.2 Packet-Optical Networking...... 9 Transport...... 40 OPTICAL NETWORKING...... 60 2.4.2 Implementing Traffic Shaping...... 31 the Home...... 69 5.5.3 Carrier Ethernet 2.0 Services...... 92 1.3 The Infinera Intelligent 4.1 Chapter Summary...... 62 4.5.2 Infinera’s Solution for Switched 5.5.4 Carrier Ethernet Service Transport Network Portfolio...... 10 2.5 Migrating Legacy TDM Services 3 OPERATING THE NETWORK...... 42 to Ethernet...... 32 4.2 Ethernet Services for Enterprises – Video Transport...... 70 Attributes...... 94 2 PACKET-OPTICAL 2.5.1 One Common Infrastructure 3.1 Chapter Summary...... 44 Business Ethernet...... 62 5.6 Carrier Ethernet Traffic 5 ETHERNET AND LAYER 2 Management...... 95 NETWORKING...... 12 for Ethernet and Legacy 3.2 Multi-layer Network 4.2.1 Serving Enterprise Customers...... 62 TECHNOLOGIES...... 72 TDM Services...... 32 Management...... 44 4.2.2 A Metro Network for Business 5.6.1 Bandwidth Profiles...... 95 2.1 Chapter Summary...... 14 2.5.2 Using the iSFP to Convert TDM 3.2.1 Management Framework...... 45 Ethernet...... 63 5.1 Chapter Summary...... 74 5.6.2 Class of Service and 2.2 The Principles of Packet-Optical Services for Ethernet Transport...... 33 3.2.2 Infinera’s Management 4.2.3 A Long-haul Network for 5.2 Ethernet Basics...... 74 Service Level Agreements...... 96 Integration...... 14 2.6 The Main Elements of an Infinera Software Suite...... 47 Business Ethernet...... 64 5.2.1 Ethernet Mode of Operation...... 74 5.7 Carrier Ethernet Operations, 2.2.1 Why Aggregate Traffic at Packet-Optical Network...... 34 Administration and Maintenance 3.2.3 Infinera Digital Network 4.3 Aggregation of IP Traffic – IP 5.2.2 Virtual LANs...... 76 Layer 2?...... 14 2.6.1 Ethernet Demarcation Units Administrator...... 48 Backhaul...... 64 (Ethernet OAM)...... 97 5.2.3 Ethernet Physical Media (PHY)...... 78 2.2.2 Ethernet Transport at Layer 2 (EDU) and Network Interface 3.2.4 A Unified Information Model for 4.3.1 IP-based Services over a 5.7.1 The Management Framework...... 97 Versus at Layer 1...... 15 Devices (NID)...... 35 Multi-layer Management...... 49 Common Infrastructure...... 64 5.3 Synchronization and Circuit 5.7.2 Standards for Ethernet OAM...... 98 Emulation Services over 2.2.3 Native Ethernet and OTN 2.6.2 Packet-Optical Transport Switches 3.2.5 Layer 2 Service Fulfillment...... 50 4.3.2 A Lean and Transport-centric 5.7.3 The Service Lifecycle...... 98 Framing...... 19 (EMXP) and the PT-Fabric...... 35 Ethernet...... 80 3.2.6 Layer 2 Service Assurance...... 50 Aggregation Network...... 65 5.7.4 Ethernet Service OAM – 2.2.4 Packet-Optical Transport 2.6.3 Optical Add-Drop Multiplexers 5.3.1 Synchronous and Asynchronous 3.2.7 Network Planning and 4.3.3 The Flexible Optical Network Transport...... 80 Performance and Fault with the XTM Series...... 20 (OADM, ROADM) and Other Brings Scalability...... 67 Management...... 100 Optical Elements...... 36 Optimization...... 51 5.3.2 Synchronization Standards...... 83 2.2.5 Packet-Optical Transport 4.4 Mobile Backhaul...... 67 5.8 Metro Ethernet Forum with the PXM...... 22 2.6.4 The DTN-X Packet-Switching 3.3 Software-defined Networking 5.4 Ethernet Protection...... 85 and Network Virtualization...... 51 4.4.1 4G/LTE and 5G Place New “Third Network” Vision...... 102 2.2.6 Leveraging the XTM Series Module (PXM)...... 36 3.3.1 Basic Concepts...... 51 Requirements on Mobile 5.4.1 Link Aggregation...... 86 and PXM in the Same Network...... 24 2.6.5 The Multi-layer Service Backhaul...... 67 5.4.2 Ethernet Ring Protection INDEX...... 104 Management System...... 37 3.3.2 Infinera’s Multi-layer SDN Vision.....56 Switching...... 87 2.7 Advantages of Packet-Optical 5.5 Carrier Ethernet Architecture Transport...... 38 and Services...... 88
4 INFINERA INFINERA 5 1 INTRODUCTION
6 INFINERA Footnote citations on page 111 INFINERA 7 INTRODUCTION
infrastructure, platform and software as a compute, storage and network functions established protocols, such as Generalized label-switched paths (LSPs) and Ethernet Client Traffic 1.1 THE CLOUD-DRIVEN Voice Transport TRANSFORMATION OF THE service (IaaS, PaaS and SaaS) offerings, is once provided from a single data center are Multiprotocol Label Switching (GMPLS), virtual connections (Figure 2). To optimize Video Technologies having a profound impact on both IT and rapidly being expanded to multiple centers but a common vision is to employ SDN the cost-to-performance ratio of Layer T, it Packet5$ NETWORK DC Interconnect Match network architectures among technology at geographically separate locations. As technology that can abstract the functions is of the utmost importance to match the Digital Over the past decade, IT infrastructure has Email Optical service providers. a result, operators need to scale, simplify of all Layer T technologies and present transport and switching technology to the Enterprise App seen a dramatic shift from local resources and increase flexibility in their transport standardized application programming traffic flow characteristics, and ultimately As a result, the traditional layered network to those located in third-party facilities. networks. Network functions virtualization interfaces (APIs) such that control can be to the end-user services. Also, these model of a hierarchical, functionally heavy Multiplex Whether we receive the services we use, (NFV) enables operators to migrate services driven by Layer C. An integrated packet- technologies are not necessarily exclusive. data network comprising several layers of including storage of our personal files, like security, voice and broadband remote optical approach to Layer T is manifested For example, Ethernet/MPLS packets may devices is evolving to a simplified model Switch streaming of personal and commercial video access server (BRAS) from dedicated in Infinera’s Intelligent Transport Network be carried in digital OTN circuits, which in of Layer C, cloud services, and Layer T, and, of course, photo albums and email, devices to software services on standard x86 architecture, which is a mix of packet, digital turn are carried by coherent optical super- intelligent transport (Figure 1). at home or on our mobile phones, they hardware within cloud data centers. NFV and optical functions that virtualize control channels as Layer T services are aggregated. all reside somewhere “on the network,” The adoption of cloud services, consisting allows service providers to drive down the at Layer C via open APIs, using both GMPLS Meanwhile, switching capabilities are or more fashionably, “in the cloud.” The of the upper OSI layers of a network, cost of network services, while delivering and SDN. essential at each of the layers in order FIGURE 2: Matching Network Technologies to acceleration of cloud services, such as is growing at a phenomenal rate. The them in a more granular and responsive to maintain the efficient use of capacity Traffic Flows manner. and provide alternative routing for traffic and OTN, but must also offer excellent 1.2 PACKET-OPTICAL engineering or protection events. Meanwhile, transport functions at the NETWORKING support for Layer 2 Ethernet functionality. Old Model New Model lower layers are converging as operators The choice of technologies to employ This functionality includes the advanced encourage vendors to develop the most Layer T traffic flows vary in many different within Layer T also has a geographical provisioning, management, verification and Firewall cost-effective network elements, giving ways, such as duration, size, time of day, aspect, splitting the industry into distinct protection of Layer 2 Ethernet services using SBC Layer C: rise to an intelligent transport layer, Layer burstiness and deterministic nature. As a sub-segments: metro/edge packet-optical L3–L7 Cloud e ices a defined group of standardized protocols B-RAS T. Layer T combines Multiprotocol Label consequence, flows can be handled by systems and long-haul/core packet-optical and procedures. Scale MPLS PE Switching (MPLS) and Ethernet packet several technologies across the multi- systems. SDN Systems at the core of a network are dealing Virtualization services with the digital functionality of layered Layer T network, including optical L2/3 Packet Metro/edge packet-optical systems are with traffic from many applications, and Optical Transport Network (OTN) switching, wavelengths, digital OTN circuits, MPLS Layer T: aimed at applications toward the edge of an as such require less service awareness. L0–L3 Layer 1 OTN nte igent Transport the capacity scaling of wavelength-division optical network. These systems typically take However, they do require the capacity to Layer 0 WDM multiplexing (WDM) and the optical path the form of WDM-based optical networking handle higher-capacity links, as well as a flexibility of reconfigurable optical add- platforms with integrated Ethernet drop multiplexer (ROADM) functionality. FIGURE 1: A Simplified Network Model Comprising Cloud Services switching. They have varying degrees of The control plane of Layer T can use (Layer C) and Intelligent Transport (Layer T) support for Layer 1 technologies such as Synchronous Digital Hierarchy (SDH)/ Synchronous Optical Networking (SONET)
8 INFINERA INFINERA 9 INTRODUCTION
METRO REGIONAL/CORE 1.3 THE INFINERA INTELLIGENT INFINERA EE F ARE I E AN NA MPLS-TP MPLS TRANSPORT NETWORK PORTFOLIO E R N A EA PBB-TE CE 2.0 MPLS-TP Infinera’s end-to-end portfolio for packet- Mobile
Layer 2 ACCESS AGGREGATION REGIONAL CORE optical networks consists of the platforms Front/Backhaul OTN SWITCHING (ODU-2) shown in Figure 4. OTN TRANSPORT, ODU-FLEX SWITCHING In the metro/edge space, Infinera is a Triple Play
Layer 1 pioneer of metro WDM through the XTM E Series, which today includes leading packet- XTG Series II 10G + Metro 100G 100G + Multicarrier Coherent optical metro access, aggregation and core Colorless, Directionless Colorless, Directionless, Contentionless Cable II NI E Gridless ROADM Overlay Gridless ROADM platforms. This platform delivers services Broadband II
Layer 0 Mixed Fiber Environment DCM Free Fiber designed for applications such as triple play broadband, cable broadband, business XTM Series DTN-X Family: XTC Series Ethernet and mobile transport. Integration of OTN + Layer 2 from edge to core without In the long-haul/core market, the DTN-X Business Ethernet technical or business compromise is the ultimate goal Family provides the world’s only commercial multi-terabit super-channel system based on Cloud Xpress Family DTN-X Family: XT Series large-scale photonic integrated circuit (PIC) Matching Network Technologies to Network Topology. FIGURE 3. technology. It includes the Packet Switching Source: IHS Markit Data Center IN ER NNE Module (PXM), which enables traffic over virtual local area networks (VLANs) or MPLS INFINERA EN F E I E RI INE E higher degree of OTN Layer 1 transport also switching across the whole chassis or LSPs with assured quality of service (QoS). In long-haul packet-optical networks, the capabilities versus Layer 2 Ethernet. This node, from any port to any other port, rather FIGURE 4. Infinera’s Intelligent Transport Network Portfolio geographical balance between Layer 1 and than just within ports on a single plug-in DTN-X family is a powerful companion to Layer 2 functionality is shown in Figure 3. unit as is common in metro/edge platforms. the XTM Series. Long-haul/core platforms typically are Long-haul/core packet-optical systems must capable of packet aggregation, OTN therefore support not only high capacity but switching, terabit-scale WDM transport and advanced ROADM functionality, all with a single control plane.
10 INFINERA INFINERA 11 2 PACKET-OPTICAL NETWORKING
12 INFINERA Footnote citations on page 111 INFINERA 13 PACKET-OPTICAL NETWORKING
2.1 CHAPTER SUMMARY The scalability and cost-effectiveness of an extra toolkit to complement Ethernet base stations and enterprise networks to Layer 1 aggregation (time or frequency Subscriber Aggregation Core Ethernet has made it the unifying service and enhance the transport capabilities more central locations. The traffic from the multiplexing), where the number of users Optical fiber provides almost lossless protocol for modern local and wide area and scalability of the network. Supervised connected equipment was transported and their data rates are fixed. Statistical L2 transmission of signals at an ultra-wide networking. Increasingly, consolidation of by a multi-layer management system that transparently to an IP core network at OSI multiplexing makes use of the fact that 1GbE range of frequencies. Packet switching, L2 the optical and Ethernet/Internet Protocol integrates the handling of OSI3 Layer 1 Layer 1, via wavelengths on fibers in ring or the information rate from each source 1GbE implemented according to the Ethernet (IP) transport infrastructures within the optical channels and Layer 2 Ethernet star topologies. As the number of endpoints varies over time and that the optical path’s 1GbE Switch family of protocols and interfaces, offers same network elements has become services, a flexible, cost-efficient and future- grew, the central IP core network had to be bandwidth only needs to be consumed 10GbE one of the most efficient ways ever to sort Switch the means to drive down both network proof telecommunications infrastructure is extended, requiring more IP routers at more when there is real information to send. 1GbE and direct streams of digital data. Packet- investment costs and the associated here and ready to be deployed. sites and with more ports, as indicated on • Statistical multiplexing in a Layer 2 optical networking fundamentally leverages operational costs. The additional support the left side of Figure 5. 1GbE This chapter deals with the principles of network also allows for the creation of new the outstanding characteristics of these for OTN1 and label-switching mechanisms transporting Ethernet traffic over WDM In this type of network, there are several types of transport services, for example 33% 100% 100% two technologies to implement a new (such as Multiprotocol Label Switching - networks and describes how these two reasons to introduce Layer 2 aggregation: services with more granular data rates and generation of telecommunication networks. Transport Profile, or MPLS-TP2) provides key technologies can be integrated in a with committed information rates that • The IP core network can be reduced in unifying architecture. It also highlights the may be temporarily exceeded. Hence the FIGURE 6: The Statistical Multiplexing of Ethernet Is size to just a few central routers instead advantages that arise from combining the operator may offer its customer a much Leveraged to “Fill the Pipes,” an Especially Useful of being spread out throughout a full Feature at the Edge of the Network Where Traffic Is two technologies, including how traffic is more differentiated service portfolio than IP Core Network IP Core Network metro network area. Layer 2 aggregation Often More Variable transported with minimal delay and without with only traditional Layer 1 transport equipment is generally less costly,4 loss of synchronization. services, as shown in Figure 6. consumes less power, has lower latency and requires less expertise to configure • Since data traffic is concentrated at Layer connectivity provides more rational traffic than IP routers. Thus, centralization of the 2 in the aggregation network, it can be Layer 2 Aggregation 2.2 THE PRINCIPLES OF PACKET- handling and reduces forwarding delay IP core network reduces the necessary handed over to IP core routers via a Layer 1 Aggregation OPTICAL INTEGRATION compared to using IP routers. (Star, Ring) investment and operational costs. few high-speed interfaces rather than 2.2.1 Why Aggregate Traffic at Layer 2? over many lower-speed interfaces. This • Layer 2 aggregation can perform The introduction of aggregation at simplifies administration and contributes statistical multiplexing of data traffic, and 2.2.2 Ethernet Transport at Layer 2 Layer 2, i.e. the Ethernet layer, brings to a lower cost per handled bit. the WDM channels of the underlying Versus at Layer 1 several benefits to network operators. optical network can be used much more • As an additional benefit, the aggregation Given the benefits of a Layer 2 aggregation Traditionally, metro aggregation networks efficiently than if only Layer 1 aggregation network itself can be used to offer network, it is important to understand how were implemented using dedicated WDM is used. Statistical multiplexing allows the services within the metro/regional area. this type of network differs from a traditional equipment that provided wavelengths bandwidth to be divided arbitrarily among For example, point-to-point Ethernet network performing Layer 1 aggregation of connecting, for example, BRAS, radio a variable number of users, in contrast to connections can be provided between Ethernet traffic over WDM. offices in a city center without loading FIGURE 5: Migrating from Layer 1 to Layer 2 Aggregation of Traffic any central router nodes. Such direct
14 INFINERA Footnote citations on page 111 INFINERA 15 PACKET-OPTICAL NETWORKING
Transporting Ethernet traffic between two sending subscriber Ethernet network and it is also unaware of any Ethernet service a very low extent will still carry them at 1 native Ethernet signal that can be handed such as 25 megabits per second (Mb/s), 200 remote sites with WDM as the underlying puts it into a digital wrapper adapted for information and can only manipulate the gigabit per second (Gb/s) as if they were over without any need for demultiplexing. Mb/s or 400 Mb/s transport services over a bearer technology can be done in two transmission over the WDM channel. At traffic at Layer 1 (Figure 7). 100% loaded. This may lead to unnecessary A Layer 2 solution will therefore normally physical 1 Gb/s port. fundamentally different ways: the receiving end, the wrapper is removed investments in Layer 1 equipment to add require a lower number of ports on the In a Layer 2 transport network (Figure Due to statistical multiplexing, a Layer 2 and the original frame is handed over to wavelengths in the WDM network. receiving switch/router, which may have • Transporting the Ethernet frames “as they 8), subscriber Ethernet networks are transport network solution is intrinsically the receiving Ethernet network. In this way, a beneficial impact on the cost of that are” (transparently) over a WDM channel, interconnected via an intermediate Ethernet In a Layer 2 transport solution, the incoming less deterministic than a Layer 1 solution. every single frame is forwarded without equipment. i.e. Layer 1 (optical) transport. switching function called the Carrier subscriber Ethernet frames are analyzed and The throughput of a Layer 2 network may modification between the two subscriber Ethernet network. A service VLAN (SVLAN) acted upon by the equipment located at the Since a Layer 2 network can use policing to suddenly change because a new service • Using an intermediate Carrier Ethernet networks. or an MPLS-TP pseudowire in the Carrier ingress point of the Carrier Ethernet network separate the effective data rate offered to a has been introduced or because the traffic network5 that in turn has its frames The Layer 1 transport solution provides Ethernet network is used to keep traffic from before being forwarded. It is possible to subscriber from the actual line rate available situation has changed. However, the Layer transported over one or more WDM a transparent path between the two each set of subscriber Ethernet networks concentrate the incoming flow of Ethernet on the access line, a Layer 2 network can 2 network can be made to behave in a channels, i.e. using Layer 2 (Carrier subscriber Ethernet networks, giving the fully separated, i.e. the SVLAN/pseudowire frames by statistical multiplexing and to also offer a more flexible and granular set deterministic way by the use of pre-defined Ethernet) transport. This is the technology highest possible QoS in terms of latency, establishes connectivity between the apply shaping and policing to the Ethernet of transport services than a Layer 1 network. capacity reservations, i.e. by the use of traffic used in a Layer 2 aggregation network. latency variation and packet loss. A Layer subscriber Ethernet networks belonging traffic. The Layer 2 solution can be made While a Layer 1 network typically only engineering. Both alternatives have advantages and 1 network is also fully deterministic and to the same subscriber. The frames of the fully “Ethernet service-aware” and can provides services at standard Ethernet line The table on the following page summarizes disadvantages. provides 100% throughput regardless of Carrier Ethernet SVLAN/pseudowire are analyze and act upon the Layer 2 traffic. rates, such as 1 Gb/s or 10 Gb/s, a Layer some of the main differences between Layer what services are carried by the Ethernet transported over channels of the WDM 2 network may offer much more flexibility, A basic Layer 1 Ethernet transport solution Multiple Ethernet signals are multiplexed 1 and Layer 2 wide area Ethernet transport. traffic. However, since a Layer 1 network network, just as described previously. takes every incoming frame from the using time-division multiplexing (TDM) is totally transparent to Ethernet traffic, A plain Layer 1 transport solution cannot in a Layer 1 solution. The aggregated concentrate the Ethernet traffic being signal is not a standard Ethernet signal, aggregated, which may result in low and consequently, demultiplexing must
Transparent transport of Ethernet frames utilization of WDM wavelengths. For be done before the original signals are Carrier Ethernet Network example, a Layer 1 network collecting handed over to a switch/router. A Layer 2 Gigabit Ethernet signals that is utilized to solution performs aggregation to a new Service VLAN
Subscriber Subscriber Subscriber Subscriber Ethernet 1 MXP Ethernet 2 Ethernet 1 Ethernet 2 MXP WDM channel (λ) EMXP EMXP WDM channel (λ) Packet-Optical Switch Packet-Optical Switch Transponder/ Transponder/ Muxponder Muxponder FIGURE 8: Layer 2 Ethernet Transport: Ethernet Traffic Is Transported via a Service VLAN in a Carrier Ethernet Network Extending Between Operator Sites FIGURE 7: Layer 1 Ethernet Transport: Ethernet Traffic Is Carried Transparently at Layer 1 over a WDM Wavelength
16 INFINERA Footnote citations on page 111 INFINERA 17 PACKET-OPTICAL NETWORKING
Layer 1 Ethernet Transport Layer 2 Ethernet Transport 2 er ice ac et ptica Sy tem • Full Layer 2 transport according to the • Plain Layer 1 transport, i.e. transponders TDM multiplexing: Collects and delivers traffic Statistical multiplexing: Collects traffic at one data without changing its format and data rate. rate and can deliver input from many sources over Metro Ethernet Forum’s (MEF) Carrier and muxponders that provide 100% one single physical interface at a higher rate. Ethernet 2.0 (CE 2.0) specifications, i.e. transparent Ethernet transport at OSI equipment that provides aggregation and Layer 1. Carrier thernet Fixed data rates: typically 100 Mb/s, 1 Gb/s, 10 Flexible selection of data rates. Allocation of concentration of Ethernet and other traffic Switch Gb/s and 100 Gb/s. bandwidth by policing and shaping of traffic. • Layer 2 transport. Using the PXM, a DTN-X with a selected set of Layer 2 functions that network supports Ethernet aggregation, Deterministic. “What goes in comes out.” Relies on traffic engineering to achieve deterministic support the transport task. Such functions behavior. switching and CE 2.0 services over an OTN include, for example, the Institute of TP core, as described in section 2.2.5. ran pon er “Unaware” of Ethernet service information and can “Ethernet service-aware.” May analyze and act on Electrical and Electronics Engineers (IEEE) TP only manipulate traffic at Layer 1. Layer 2 control information, i.e. can be used to 802.3ad Ethernet Link Aggregation, traffic TP create Ethernet transport services with different characteristics and quality of service. shaping, policing and bandwidth profiles with guaranteed bandwidth allocation. The Lowest delay, jitter and packet loss. Statistical multiplexing implies a risk for delay, jitter XTM Series’ Layer 2 network elements – the and lost packets. 2.2.3 Native Ethernet and OTN Framing ine Sy tem rop EMXP packet-optical transport switches – The Carrier Ethernet network of a Layer 2 mp ifier Layer 1 performance management based on bit Layer 2 performance management based on VLANs are also Layer 1-aware, meaning that they transport network is built on WDM optical errors (cyclic redundancy check or CRC). and Ethernet Virtual Connections, latency, jitter, can be connected directly to a WDM link frame loss, etc. and support features such as forward error channels, i.e. the frames of the Carrier Embedded management channels via overhead Embedded management channels via separate correction (FEC) at the optical layer. Ethernet network are transported by the bytes in line signal wrapper. management VLAN at Layer 2. underlying WDM optical system. The FIGURE 10: The Primary Entities Forming a Node in an XTM Series Packet-Optical Network All the traffic units mentioned above are of Ethernet payload data can be packaged the plug-in type and can be installed side FIGURE 9. Characteristics of Layer 1 and Layer 2 Wide Area Ethernet Transport in the transport containers of the optical by side with optical units such as ROADMs in the XTM Series platform’s different system in several different ways. Optical OTN is a more recent addition to the It supports FEC and management functions The XTM Series provides the operator with Gigabit Ethernet connection is utilized. This chassis options. As an example, Layer transport standards/systems such as SDH, standards for public telecommunications for monitoring a connection end-to-end three principal alternatives for transport of enables the network operator to analyze the 2-capable EMXPs can be used at the edge SONET and OTN therefore incorporate networks and is sometimes referred to over multiple optical transport segments. Ethernet traffic: of the network to collect and aggregate various adaptation methods, e.g. Generic by its International Telecommunication wavelength utilization and avoid unnecessary OTN wraps any client signal in overhead investment in transponders/muxponders Ethernet traffic and hand it over to Layer 1 Framing Procedure (GFP), to allow for Union Telecommunication Standardization • Plain Layer 1 transport, i.e. transponders and information for operations, administration to launch additional wavelengths. Another transponders or muxponders that provide packet data transport. These methods Sector (ITU-T) name, G.709. The standard is muxponders that provide 100% transparent and management. The client signal to differentiating XTM Series feature is the continued transport with cost-efficiency and a allow the mapping of variable-length, designed as a “digital wrapper technology” Ethernet transport at OSI Layer 1. be transported is mapped into an optical ability to inject/extract management VLANs high quality of service (Figure 10). higher-layer client signals, such as a flow to transport both packet-mode traffic, • Ethernet-aware Layer 1 transport, i.e. on a Gigabit Ethernet client port. This enables channel payload unit (OPU). The OPU is The DTN-X Family with the PXM provides of Ethernet frames, to the channels of the such as IP and Ethernet, and legacy SDH/ muxponders that provide 100% transparent direct remote management of Layer 2 devices then encapsulated into the basic unit of a similar set of choices for the transport of WDM network. SONET traffic over fiber optics with WDM. transport but have Layer 2 features, such as via the same data communications network Ethernet traffic: information transport by the protocol, the providing information on to what extent a (DCN6) solution as for the optical transport equipment.
18 INFINERA Footnote citations on page 111 INFINERA 19 PACKET-OPTICAL NETWORKING
Header SVLAN User data categories: transport systems and switches. Signal Data Rate Application Examples Native Ethernet means that no additional ID Carrier Ethernet Network frame OTN transport systems follow OTN framing is applied to the data payload at ODU0 1.24416 Gb/s Transport of a timing-transparent transcoded (compressed) standards when framing transported signals 1000BASE-X signal or a stream of packets (such as Ethernet, MPLS the edge of the network. By treating the for a specific route. The transport systems or IP) using Generic Framing Procedure. Ethernet packets natively, it is possible OPU Carrier Ethernet frame are therefore compliant with the sections of to inspect them within the intermediate Overhead Optical Channel Payload Unit (OPU) ODUflex Any configurable rate Transport of a stream of packets (such as Ethernet, MPLS or IP) the OTN standards relating to transmission, (GFP) using Generic Framing Procedure. network nodes and to act upon the Ethernet such as framing, FEC, performance ODU2e 10.3995253164557 Gb/s Transport of a 10 Gigabit Ethernet signal or a timing-transparent headers so that the combined benefits of monitoring, etc. Transport systems use transcoded (compressed) 10 Gigabit Fibre Channel (10GFC) signal. Layer 2 intelligence and efficient Layer 1 ODU Payload transponders or muxponders to terminate transport can be realized. This becomes Overhead Optical Channel Data Unit (ODU) the signal. OTN switches take this one stage FIGURE 12: Characteristics of Some of the Optical Channel Data Units especially important at the edge of the further and add an OTN switching fabric that is capable of demultiplexing incoming OTU Payload Native Ethernet Overhead Optical Channel Transport Unit (OTU) OTN signals and switching the lower-speed control information, i.e. to perform • Data is sent as ordinary signal components to other high-speed statistical multiplexing and differentiation Ethernet frames. • Access to Ethernet interfaces. of services at the Ethernet level. control information at intermediate nodes. FIGURE 11: OTN Signal Structure and Terminology. The Carrier Ethernet Frame is Carried as the Payload • Framing according to the OTN of an Optical Channel Payload Unit (OPU) OTN framing standard: The frames of the Carrier • Data is packed into a 2.2.4 Packet-Optical Transport Ethernet network are carried over a WDM wrapper for transport via an OTN network. with the XTM Series wavelength by ODUs according to the OTN • Ethernet control The XTM Series supports two types of standard. This encapsulation is especially information is optical channel data unit (ODU), which is that optical channels may be underutilized if encapsulated. encapsulation of Carrier Ethernet traffic for favorable when the Carrier Ethernet carried within an optical channel transport an efficient multiplexing scheme is not used WDM transport: network extends over longer distances, unit (OTU), defining the line rate of the at the ODU level. Three ODU multiplexing or the data is to be transported via an connection (Figure 11). schemes important in a packet-optical intermediate core OTN network. Using Carrier Ethernet Network network are ODU0, ODUflex (GFP) and ODU2e framing between EMXPs allows OTN includes OTU standards for a fixed • Native Ethernet: The frames of the Carrier the use of OTN’s inherent FEC mechanisms ODU2e. ODU0 units are optimized to carry Ethernet network are transported as is, i.e. number of line rates covering the range and optical path monitoring bits, which 1 GbE signals within an ODU1 unit, which using the same preamble and end-of-frame Subscriber Subscriber from 2.6 Gb/s up to 112 Gb/s. These are important for long-reach links. Since Ethernet 1 Ethernet 2 in turn is carried by an OTU1 transport unit. sequence as on an ordinary LAN7 when bitrates have been selected based on the the encapsulation is in ODU format, these EMXP EMXP ODUflex units can be multiplexed into any forwarded over the WDM wavelength. This WDM channel (λ) overall requirements of the optical transport data units can also easily traverse any of the higher-rate ODUs and carried at the is the standard encapsulation recommended Packet-Optical Switch Packet-Optical Switch intermediate OTN switches transparently system, but are not necessarily multiples of by Infinera for metro/aggregation networks, appropriate OTU channel line rate (Figure 12). before reaching their final destination the bitrates of the signals to transport, e.g. since it allows each intermediate node (Figure 13). FIGURE 13: Native Ethernet and OTN Framing with XTM Series EMXPs. The Two-colored Bars Symbolize Gigabit Ethernet (GbE). The consequence is OTN systems are often divided into two to examine and manipulate the Ethernet Ethernet Frames, with the Control Information in Red
20 INFINERA Footnote citations on page 111 INFINERA 21 PACKET-OPTICAL NETWORKING
High priority Ethernet frame to be mapped into an ODU2e Native Ethernet OTN Framing and can range from 1.25 Gb/s to 100 Gb/s, Ethernet services and Layer 2 QoS. Instead of Business apps data unit ready for transport into the OTN allowing for right-sized transport and core, including OTN switches. This is very Ethernet frames are Ethernet frames are requiring a dedicated port for every service, maximum bandwidth efficiency (Figure 16). transported natively transported in OPUs the PXM supports port consolidation, which Low priority useful once the traffic has been aggregated over WDM channels within ODU and OTU can provide multiple services or flows on one Having the packet streams mapped to Signaling Interactive as much as possible to ensure the best with minimal extra containers according to voice, video overhead. the OTN standard. port, overcoming the requirement of one standard ODUflex containers allows for utilization of the 10 Gb/s circuit. Figure 15 physical port and fiber connection per service. multiplexing with other types of ODU contrasts the differing characteristics of native The service information The Ethernet frame Loss intolerant contained in the is wrapped inside containers and for switching the flows as data Ethernet and OTN framing. Ethernet frame can the OPU/ODU and be accessed at every cannot be read and appropriate at the optical layer using the An OTN core network can also provide a node of the network. acted upon without The PXM maps incoming streams of data DTN-X OTN Switch Module (OXM). Hence, Loss tolerant Operator unified transport layer where core nodes com- This allows for Layer 2 demultiplexing. packets, such as Ethernet VLANs, into it becomes possible to apply standard Layer data service differentiation flows that can be co-routed through the bine traffic from OTN muxponder-based Lay- at intermediate network 1 traffic engineering and protection to flows er 1 services with EMXP-based, ODU-framed nodes. network with other flows that have the of Ethernet packets as well. same destination. The packet flows with the Ethernet services. End-to-end performance No FEC without an Includes FEC and same destination are groomed into a single FIGURE 14: Service-aware Transport Enables a Differentiated Service Offering with Multiple Classes of Services monitoring is achievable even over multi- additional Layer 1 optical path monitoring transponder. mechanisms, which ODUflex container for efficient routing. with Different Characteristics ple operators’ networks through OTN with are important for long- tandem connection monitoring and Carrier distance links. ODUflex containers are built as necessary Ethernet’s inherent operations, administration Traffic can be carried Provides direct and maintenance (OAM) capabilities. “transparently” over an compatibility with N x 10 GbE network where decisions about traffic For example, frames containing data for OTN network. intermediate core prioritization are made and where traffic high-value and high-quality IP services OTN networks – ODU containers can be is aggregated to “fill the pipes.” The such as IP/MPLS, IP-virtual private network 2.2.5 Packet-Optical Transport with the PXM multiplexed and Consolidate 10G ports wrapping of traffic into full OTN can then be (IP-VPN) or virtual private LAN service (VPLS) switched like any other into 100G ports flow. done at the handover to the core network, can be switched to paths for transport to the Packet-aware The DTN-X Family of platforms performs ex OTN Core after aggregated pipes of traffic that are desired IP devices. By contrast, frames that Suitable primarily for Applicable in most U integrated WDM transport and OTN OD correctly shaped have been created, are destined for transport services (Ethernet, metro and regional situations but especially ODU ex switching, grooming and service protection at networks where traffic in core and long- 100 GbE N x 10 GbE avoiding wasted bandwidth. MPLS-TP or OTN) can be kept within the Layer 1, based on large-scale PIC technology. is aggregated and distance networks OD U optical transport network, with minimal Using the PXM, the DTN-X Family can map service differentiation is where traffic has been ex Native Ethernet uses the VLAN tag or MPLS applied. aggregated into larger use of expensive Ethernet switching and IP Ethernet VLANs to ODUflex subnetwork label to switch the frames to ports associated streams. routing resources (Figure 14). connections, which enables ODUflex-based with either IP services or transport services. FIGURE 16: The DTN-X Family with traffic engineering in long-haul networks. Some Characteristics of Native Ethernet Each of these service domains is optimized The EMXPs have OTN framing capabilities for FIGURE 15: Integrated PXM Consolidates Framing and OTN Framing Client Ports and Routes Packet N x 10 GbE and simplified for particular service types. ODU2e, including optional G.709 FEC on all With the introduction of the PXM, the DTN-X Family supports statistical multiplexing for Flows over ODUflex Subnetwork 10 Gb/s ports. The optional ODU2e framing Connections on 10 Gb/s ports allows the native 10 Gb/s
22 INFINERA Footnote citations on page 111 INFINERA 23 PACKET-OPTICAL NETWORKING
The flexible architecture of the packet PXM OXM DWDM-LM transport network allows for cost-efficient Core OTN
OTN aggregation of all types of packet-based Packet XTC-4/10 Processing OTN traffic. Service routers can be consolidated OTN Client to a few central sites and equipped with OTN (e.g: 100GbE) ODUj ODUj ODUj N x ODUj Super-channel high-speed interfaces, thereby reducing (e.g. ODUFlex (e.g. ODU2) (e.g. ODU2) (e.g. ODU2) (e.g. 5x or ODU2e) ODU4) overall network investment and operational PXM costs. The packet services offered adhere to PXM recognized Metro Ethernet Forum standards Client Service Support OTN Network Interfaces Network Service Support FIGURE 17: Mapping of Ethernet Traffic into ODUflex Containers Allows for Multiplexing and Switching at Layer 1 and can be managed according to well- Using •Standard Untagged OTN Ethernet Principles • ODU exi SNCs • MPLS-TP (Ethernet PWE) Metro 1 • Priority Tagged Ethernet • ODU2e SNCs* • 802.1Q (C-VLAN) Encapsulation*established OAM procedures defined PXM • 802.1Q (C-VLAN) Encapsulation • 802.1ad (C & S-VLAN) Encapsulation* SGW for Carrier Ethernet. Layer 1 protection CATV & Mapping• 802.1ad the (C Ethernet & S-VLAN) traffic Encapsulation to the ODU • MPLS-TP LSP* schemes can be applied in the core network Internet • MPLS-TP LSP* services, such as those defined by Metro Aggregation can be done in various ways, for example segment, while additional Layer 2 protection Ethernet Forum for CE 2.0, can be offered to Metro 2 Metro .N by carrying the Ethernet VLANs in MPLS- users in one or more geographical regions. schemes, such as MPLS-TP switching CCAP TP pseudowires (PWs), creating additional As shown by the dotted lines in Figure 19, and G.8032v2 Ethernet Ring Protection Business MSAN XTC-2 options for traffic engineering and packet services may span one or more metro Switching (ERPS), can be applied when Internet protection (Figures 17 and 18).8 networks, have their endpoints in an EMXP or services are delivered via EMXPs. NID be delivered directly to, for example, a service NID router from a node in the core network via the PXM in the DTN-X. 2.2.6 Leveraging the XTM Series and PXM in Mobile the Same Network Backhaul Packet Processing Steps: Hub • Incoming VLAN examined for The XTM Series, with its advanced handling Cell site destination router of Ethernet services, can seamlessly interwork VLAN 1 to A MPLS-TP ODU2e to A • VLANs mapped to with the PXM and its super-channel-based pseudowires (PWs) and then to transport of Ethernet traffic mapped to ODU Ethernet ODU ex or ODU2e* OTN VLAN 2 to B PWE X MPLS-TP FIGURE 19: Multiple Metro Networks Interconnected by a High-capacity OTN Core. Each Dotted Line Represents an Ethernet containers. Deploying a high-capacity OTN ODU ex to B containers • Multiple VLANs per Service Carried over the Packet-Optical Network VLAN 3 to B core based on the DTN-X Family with PXM PWE Y MPLS-TP ODU2e/ODU ex allows for the interconnection of multiple ODU ex is resized on demand metro networks into one intelligent packet ODU ex (hitless resizing) transport infrastructure, as depicted in Figure Hitless resizing FIGURE 18: Adaptation of 19. In this infrastructure, end-to-end Ethernet Ethernet Services into ODU Containers (Example)
24 INFINERA Footnote citations on page 111 INFINERA 25 PACKET-OPTICAL NETWORKING
In summary: In the access and aggregation MPLS-TP label-switched path (LSP) the MPLS-TP network. Using MPLS-TP Both the tunnel and the LSP can be Ethernet frame from user part of the transport network, where service terminology, the entry and exit nodes are envisaged as predefined circuits for Ethernet Payload granularity is required, a service-aware referred to as MPLS-TP provider edge information to follow through the network, header packet-optical mechanism is beneficial Carrier Ethernet Network (PE) nodes, and any intermediate nodes and consequently tunnels and LSPs are to support different QoS. Also, access being passed by the LSP are referred to configured in advance from the network to Ethernet OAM bytes and service tags as MPLS-TP provider (P) nodes. Often the management system. A key feature of MPLS labels enables end-to-end management of physical node performing the PE function MPLS-TP that distinguishes it from classic used for Subscriber Subscriber 9 switching Ethernet services. Using native Ethernet Ethernet 1 Ethernet 2 is called a label edge router (LER) and the IP/MPLS is that management framing offers benefits from revenue- EMXP EMXP intermediate transit node is called a label and protection are designed Ethernet MPLS PW Pseudowire payload generation, investment and operational WDM channel (λ) switching router (LSR). EMXPs can act as to operate without a dynamic header label label Packet-Optical Switch Packet-Optical Switch perspectives in the access and aggregation both an LER and an LSR, and may also control plane, i.e. similar to parts of the network. On the other hand, combine these roles (Figure 21). a traditional SDH/SONET Local Ethernet header Forwarded Ethernet frame OTN has all the benefits of a long-haul FIGURE 20: Using MPLS-TP to Define Label-switched Paths Within the Carrier Ethernet Network network, where circuits are set An MPLS-TP tunnel is a pre-defined optical transport network once the traffic has up by the management system. MPLS-TP transport path from the source FIGURE 24: Ethernet over MPLS Encapsulation been sufficiently aggregated and traverses LER to the destination LER. The MPLS-TP The actual data traffic is carried by a the core part of the network. Combining These can be addressed by the use of are switched based on their SVLAN tags and tunnel always has an active LSP that defines pseudowire inside the LSP/tunnel. One these two framing schemes within one MPLS-TP, which is the transport profile of media access control (MAC) addresses. the primary and working paths. It may MPLS-TP LSP may carry one or more A pseudowire is an emulation of a Layer 2, intelligent packet-optical transport multi-protocol label switching. also have a protection LSP that defines a pseudowires, i.e. the pseudowires offer a point-to-point, connection-oriented service architecture makes it possible to exploit the A label-switched path is defined between recovery path (Figure 22). means for multiplexing of traffic (Figure 23). over a packet-switching network (PSN), benefits of both technologies in a common MPLS-TP is a way to simplify Carrier nodes where traffic enters and leaves Ethernet services by pre-defining from one attachment circuit (AC) to another. infrastructure for service creation. Label Switching Router The pseudowire used in MPLS-TP is a connection-oriented services over packet- (LSR) LSR LSR Pseudowire MPLS-TP Tunnel/LSP based networking technologies in a way Active LSP MPLS-TP Tunnel connection established between two MPLS- that gives support for traditional transport TP LERs across the MPLS-TP tunnel/LSP with Label Edge Router LER LER 2.3 MPLS-TP FOR TRAFFIC LSP 1 the attachment circuit frames encapsulated operational models. It takes the advantages (LER) B B ENGINEERING AND SERVICE B as MPLS data (Figure 25, next page). SCALABILITY of MPLS concepts by adding more flexibility and network manageability than the basic A C A C A C While Carrier Ethernet networks and the use Ethernet SVLAN architecture (Figure 20). of service VLANs bring great advantages to LER LER LER Attachment Attachment Circuit (AC) the packet-optical network, there are some LSP 3 D Label Switched Path 2 D Protect LSP Circuit (AC) D (LSP 2) limitations in terms of protection options, Principles of MPLS-TP FIGURE 21: MPLS- FIGURE 22: Data LER + LSR LSR FIGURE 23: LSR traffic engineering and service scalability. MPLS is a technique that forwards packets TP Framework MPLS-TP Is Carried by a based on labels, as opposed to a standard Tunnel Pseudowire Defined Carrier Ethernet network where the frames Within the MPLS-TP Tunnel and LSP
26 INFINERA Footnote citations on page 111 INFINERA 27 PACKET-OPTICAL NETWORKING
which bring a mesh-based structure Applications layer - Tunnel PW MPLS-TP is fully supported by the EMXP to the wavelength routing and paths Applications Layer 7 range. Any physical port on an EMXP can Tunnel ID available through the physical network in/out support both native Ethernet and MPLS- AC pw-label AC TDM Ethernet Client service layer TP, allowing operators to deploy MPLS-TP for any services. One previous drawback Edge out label in label Transit out label in label Edge with Ethernet was that it was not well- Active in label out label in label out label Active where and when it makes sense for them. It is Pseudowire possible to run MPLS selectively per port or to suited for protection and restoration over LSP I/F I/F I/F I/F MPLS-TP layers LSP ID separate MPLS traffic based on MAC address mesh-based networks, as the available Label Switched (LSP)/Tunnel and VLANs within the same port. This allows protection schemes were largely based seamless migration and coexistence with the on point-to-point or ring architectures. Ethernet Link layer LER LSR LER two protocols running independently side by MPLS-TP is highly suited for a mesh-based side. environment and allows network operators ODU framing FIGURE 25: Relation Between Pseudowire, Tunnel and LSP to design network resilience strategies that Physical layers - WDM Layer 0/1 are closely aligned to the physical structure WDM Services can be extended over DTN-X- of the network, ensuring the best possible based networks through the PXM, which resilience and service uptime. A Layer 2 transport service is established Both Ethernet SVLAN and MPLS-TP supports MPLS-TP and VPLS over ODUflex in an OTN core network. FIGURE 26: MPLS-TP OSI Network Layers. The Two Variants of the Physical Layer Correspond to Native between two attachment circuits and the forwarding techniques have their own Ethernet Framing and OTN Framing Respectively service is carried by a pseudowire. The benefits, and it is often advantageous to In order to deploy MPLS-TP in a production MPLS-TP—Easy Service Creation pseudowire travels through the network in be able to offer services based on both network, it can be introduced either as Another advantage of MPLS-TP is that it an MPLS-TP tunnel. The MPLS-TP tunnel technologies. For example, multicast an overlay to the existing Ethernet or breaks service creation into two steps. First, to the network. Second, it brings a very network. Since each subscriber network is in turn mapped to at least one LSP: the services can often be deployed directly over incrementally as an evolutionary build-out in tunnels are created between endpoints familiar look and feel to service creation may include an extensive number of active LSP. SVLANs on Ethernet, whereas point-to-point parallel on the same networking hardware. within the network for service and protection as it is similar to the processes involved in devices and MAC addresses, this results trunks requiring protection can benefit Figure 26 on the next page summarizes how Infinera believes in a smooth evolution, paths for MPLS-TP-based services. Then traditional transport networks. This helps in a need for large MAC address tables more from the MPLS-TP features. However, transport in an MPLS-TP network relates to provided by a “ships in the night” capability, the network administrator simply creates operators migrate from traditional transport in each network node, creating various MPLS-TP pseudowires can also be used for the OSI model. Note that Ethernet framing with Ethernet and MPLS-TP acting in parallel new services by adding them to the tunnel networks to packet-optical networks. problems and extra equipment cost. Using multi-point communication using VPLS,10 is present at both the link layer and the as independent non-interfering protocols in endpoints as pseudowires, safe in the MPLS-TP, the subscriber MAC addresses which allows geographically dispersed sites client service layer. the same system. knowledge that all routing aspects of the are encapsulated within the pseudowire to share an Ethernet broadcast domain by service have already been handled. This payload and not seen by the intermediate connecting sites through pseudowires. MPLS-TP Resolves the brings two distinct advantages. First, it MAC Scalability Problem switches of the Carrier Ethernet network. MPLS-TP in ROADM-based Optical Networks makes the solution more scalable, as it is In an SVLAN-based Carrier Ethernet Packet-optical networks are often deployed simpler to add a large number of services network, all MAC addresses in the attached today over ROADM-based optical networks, subscriber Ethernet networks are visible to every switch within the Carrier Ethernet
28 INFINERA Footnote citations on page 111 INFINERA 29 PACKET-OPTICAL NETWORKING
2.4 TRAFFIC MANAGEMENT shaping is typically done per port, interface or other physical entity, while policing • The EMXP has a QoS Engine with 8 Queues Layer 2 networks rely on packet switching High ∙ Requires buffering + scheduling can be applied more granularly and per Transmit Outgoing where different CoS are mapped against ∙ Delays some packets and statistical multiplexing when forwarding queue packets Ethernet service, VLAN, etc. Incoming packets • Queues are emptied by a scheduler that information. Statistical multiplexing Medium Classify can operate in different modes Infrastructure level (per port) time time implies that the ports of the network can As shown in Figure 27, policing plays an Normal Before shaping After shaping ance ie a c ica is se o • Strict priority be allocated more “bandwidth” than is important role in traffic management of a ance an ai o e s sc iption o • Round Robin se ices o te totally available, so a Layer 2 packet-optical Carrier Ethernet services and their individual Low SLAs, as described in section 5.6. • Weighted RR network must have an inherent mechanism Length de ned • While a frame is waiting to be scheduled out ∙ Requires no buffering to control the amount of information by queue limit ∙ Bandwidth is metered Traffic shaping is done by imposing an on the line it is buffered entering the network and passing additional delay on some packets such that over network links. These procedures • If there’s no buffer left the frame is dropped time time Service Level the traffic conforms to a given bandwidth Before policing After policing are collectively referred to as traffic • This is done in wirespeed profile. Traffic shaping provides a means Classification by: Queueing * In real implementations traf c is measured with leaky-bucket to allow controlled bursts management, and the two most important Interface hardware to control the volume of traffic being sent • Protocol buffer resources • Ethernet such tools are traffic shaping and policing. out on an interface in a specified period • Source interface (bandwidth throttling), and the maximum • VLAN FIGURE 27: The Difference Between Traffic Shaping and Traffic Policing • Prio-bits rate at which the traffic is sent (rate limiting). 2.4.1 Traffic Shaping and Policing Traffic shaping is a traffic management A drawback with traffic shaping is increased FIGURE 28: Traffic Shaping: Quality of Service Is Assured by Buffered Queues and Scheduled Forwarding of Packets latency and jitter for the Ethernet Virtual These switches do not have to be designed network, there is no such upper limit for the technique that delays some frames in a Connection (EVC), but the gain can be with the number of subscriber Ethernet number of subscribers that can be handled packet-optical network node in order to better throughout since the overall flow MAC addresses in mind. by a network using MPLS-TP. bring them into compliance with a desired 2.4.2 Implementing Traffic Shaping When a network element is congested, it traffic profile. Traffic shaping is a form of of frames may be improved: instead of starts to drop frames if there is no buffer Of course, as the MPLS-TP services rate limiting, as opposed to the policing of dropping traffic in a policer, it may be better Traffic shaping is typically done by assigning left to allocate. Weighted random early supported by Infinera’s XTM Series to shape the traffic to make sure no frames incoming packets of different priority to MPLS-TP Allows for a Virtually Unlimited bandwidth profiles, where excess frames are detection (WRED) is a way for a network are delivered over the same hardware are lost (or at least as few as possible), separate queues in the network element. Number of Customers simply dropped. Normally, traffic shaping element to ask clients to require fewer platform as native Ethernet services, they avoiding retransmissions at higher protocol For example, the EMXP uses eight queues The IEEE 802.1Q standard allows for a is not part of any subscriber service level resources from the network. However, also benefit from the same transport- layers. for different classes of services (CoS). The maximum of 4094 SVLANs in a Carrier agreement (SLA), but rather an internal WRED only acts on traffic that is transported like performance with extremely low queues are emptied by a scheduler that can Ethernet network, with one SVLAN typically network mechanism used by the operator to operate in different modes, e.g. strict priority by the TCP/IP protocol that ensures that latency and almost zero jitter, and can be required per subscribing customer. Since “even out” traffic flows and create fairness or round robin. While a frame is waiting to be data is re-sent. With WRED, different discard combined with synchronization schemes MPLS-TP uses tunnels and label-switched between users of network resources. Traffic scheduled for transmission out on the line, it profiles can be set for different CoS, and such as Synchronous Ethernet (SyncE) when paths to define the connectivity within the is buffered in the element, and if there is no TCP traffic can be dropped randomly so required, in mobile backhaul networks buffer left the frame is dropped (Figure 28). that the overall load on the network is for example.
30 INFINERA Footnote citations on page 111 INFINERA 31 • Provides TDM services at the edge of an Ethernet network Transparent transport of: • Transparent TDM over packet (TSOP and SaTOP) a oa st ct es are outlines in open IETF drafts/standards* e ea tes an p otection p otoco s nc oni ation • TDM streams are “chopped up” and carried in ine ates 1 1 an 1 Ethernet packets PACKET-OPTICAL NETWORKING
decreased. Dropping traffic at random bandwidth perspective, Ethernet traffic has approach where the new packet-mode Steady stream of bits prevents a “global TCP sync,” in which all already surpassed the amount of legacy network also provides legacy TDM services • Layer 1 services such as E1, STM-1/OC3, TCP sessions try to increase their bandwidth TDM traffic. Operators are faced with the is an attractive alternative (Figure 29). channelized STM-1/OC3, STM-4/OC12 and TDM Network at the same time, creating a “perfect challenging task of maintaining existing STM-16/OC48 can easily be adapted for Many legacy TDM services can be replaced transport over Ethernet through the use of storm.” TDM services while upgrading networks by Ethernet equivalents and are well- Infinera’s Intelligent SFP (iSFP) pluggable for new Ethernet services, all in the most SDH: 810 bytes Encapsulation Ethernet suited for the packet-optical infrastructure optics that provide circuit emulation over headers headers Ethernet Network cost-efficient way. More capacity is required E1: 256 bytes described in the previous sections. However, SyncE-capable Ethernet. The iSFP modules 2.5 MIGRATING LEGACY TDM to cater to the growing amount of packet- fit into any EMXP Gigabit Ethernet port, there is a significant portion of traffic that Steady stream of Ethernet frames SERVICES TO ETHERNET mode traffic generated by Internet access, allowing a very flexible and tactical service cannot simply be migrated to Ethernet. video on demand and cloud computing, migration. 2.5.1 One Common Infrastructure for This fact creates a dilemma for network FIGURE 30: Transparent SDH/SONET and E1 over Packet while the shrinking amount of TDM traffic * RFC 4553 and draft-manhoudt-pwe3-tsop Ethernet and Legacy TDM Services operators as SDH/SONET systems start to must be taken care of to sustain revenues Transmission systems based on Layer reach end of life, or if a large SDH/SONET from existing services. Building a separate 2.5.2 Using the iSFP to Convert TDM Services 1 TDM technology, such as SDH and or even E11 network must be supported for new infrastructure for Ethernet traffic while for Ethernet Transport The TDM traffic is mapped into an Ethernet SONET, have been the basis for many a small number of services or subscribers. maintaining a shrinking SDH/SONET Virtual Connection that can be transported In Infinera’s iSFP solution, synchronization telecommunication services offered during network is one option, but an integrated The iSFP is used to deliver TDM services over either as an Ethernet service VLAN or via an transport is provided by the differential clock the last few decades. However, from a recovery (DCR) mechanism, transferring a network built for Carrier Ethernet services. MPLS-TP service. The EMXPs can also be used Infinera’s packet-optical platforms offer the SDH/SONET clock to the transparent It provides circuit emulation for STM-1/OC3, to perform Layer 2 aggregation to ensure full several alternatives for the handling of legacy STM-4/OC12 or STM-16/OC48 SDH/SONET emulated service with SyncE as a reference at utilization of higher-speed 10 Gb/s Ethernet Layer 1 CE 2.0 Services TDM-based services in a way that greatly services as well as E1 services or channelized both ends. DCR transfers the synchronization Layer 1 Point-to-Point Services ine facilitates a gradual shift toward one common, connections. The adaptation of TDM services clock to the emulated service and extracts it Point-to-Point Services STM-1 signals over a SyncE-capable Carrier 1 to Ethernet follows open Internet Engineering again at the far end of the service. (See also 1 1 ee packet-mode, Ethernet infrastructure for all Ethernet network. 1 Existing Layer 2 Services 1 ccess traffic. Task Force (IETF) standards (Transparent SDH/ section 5.3.2.) 1 ine 1 ansit The iSFP module performs packetizing of SONET over Packet and Structure-Agnostic 1 est e o t t anspo t • Existing Layer 1 services carried over the TDM service, converting for example an Time Division Multiplexing over Packet) for SDH/SONET or E1 networks can be STM1/OC-3 service into a 170 Mb/s Ethernet transported separately from Ethernet stream or an STM-4/OC-12 service into a TDM-over-packet transport (Figure 30). The two directions of the service can operate as two independent timing services using different wavelengths, while 680 Mb/s stream, creating a transparent bit Ethernet/MPLS-TP Transport still using the same WDM platform and domains or one direction can be frequency- SDH/SONET Ethernet CE 2.0 pipe between two locations in the packet- SyncE Capable optical transmission network. Infinera’s optical network. The existing TDM service, Synchronization locked to the other direction. platforms include powerful ROADMs for including payload structures, overhead bytes Network synchronization is the cornerstone In cooperation with an ecosystem partner, flexible handling of the optical paths and and protection protocols, is transparently of an SDH/SONET network. Here the quality Infinera also supports synchronization muxponders/transponders that adapt SDH/ migrated to the Carrier Ethernet network. of the underlying Ethernet network is critical. SONET traffic to optical transport. FIGURE 29: Migration Toward a Single, Packet-oriented Transport Infrastructure for All Services Service adaptation is done at the edge of the The Infinera XTM Series provides excellent network, and standard Ethernet frames are Synchronous Ethernet performance due to used between the ingress and egress points. patented innovations in circuit design.
32 INFINERA Footnote citations on page 111 INFINERA 33 PACKET-OPTICAL NETWORKING
Primary Reference Clock how these functions are distributed in a A third central objective when developing 2.6.2 Packet-Optical Transport Switches complete network and between products. the Infinera packet-optical portfolio has Infinera’s Ethernet Demarcation Unit (EMXP) and the PT-Fabric been to enable multi-layer traffic and service is an independent unit that supports A key objective in the development of the management. Depending on traffic load or service level agreement management In the metro aggregation network, traffic Infinera packet-optical portfolio has been capabilities, including sophisticated traffic link degradation, the packet-optical network enters via an EMXP in the first node. The same to expand the number of services that management and hierarchical quality of must be able to switch between different physical node may also include muxponders/ can be provided by and over an optical service mechanisms; standard end-to-end transponders using additional WDM channels 1 transport alternatives, in order to deliver operations, administration and maintenance 1 infrastructure. More services over the same for fully transparent Layer 1 transport, 1 1 the highest possible quality of service for and performance monitoring and extensive 1 with Layer 2 transport used only where 1 network mean more revenue and less cost 1 subscribers. The tight integration between fault management and diagnostics, all to 1 for the operator. By integrating a selected inspection and OAM information is required at Layer 2 and Layer 1 functionality in the reduce service provider operating costs and set of Layer 2 functions with the optical layer, intermediate points, and where traffic needs o on ti ing e e ence o nc ono s t e net architecture also enables software-defined capital expenses. A Z the network becomes much more potent to be aggregated by statistical multiplexing. i e entia c oc eco e networks and advanced traffic management. in terms of service offerings and can be The same set of demarcation functionality is The EMXPs are available in various made more scalable. The tight integration also available via Infinera’s Network Interface configurations and mounting options, such FIGURE 31: Using Synchronous Ethernet (SyncE) as a Reference for SDH/SONET Differential Clock Recovery Device, which is a port device supported by between Layer 1 and Layer 2 also makes it as one- or two-slot chassis-mount units and 2.6.1 Ethernet Demarcation Units and Network its parent EMXP. The NID performs service possible to increase resilience and improve a temperature-hardened access unit for Interface Devices OAM but leaves the service policing and traffic management in ways not possible deployment in non-controlled environments – tagging to be done by the EMXP, reducing all with full wire-speed forwarding on all ports management and monitoring. Probes can schemes are replaced by MPLS-TP and with less integrated approaches. Demarcation of a provided service is a key the cost and complexity of the customer- for all traffic types. be deployed in the network to monitor Ethernet protection options. Protection is located equipment. Another main objective of the chosen function of the packet-optical access network, SyncE quality, which is used as a reference supported by MPLS-TP for topologies such packet-optical architecture has been to as it enables the service provider to extend for both ends of the transparent TDM as ring, mesh and partial mesh. In addition, make it agnostic to the underlying transport its control over the entire service path, service. The probes can also be used to protection of the employed Ethernet Virtual starting from the customer handoff points. network technology. Traffic flows over a monitor the SDH/SONET sync quality of the Connections can be accomplished by the The customer’s equipment is connected to the ROADM-based, photonic network that connected systems to provide confirmation Ethernet Ring Protection Switching (ERPS) Carrier Ethernet network via a provider-owned provides connectivity and can be used for FIGURE 33. Two Examples of of the sync in the TDM service that is carried and Link Aggregation Group (LAG) options, demarcation device (Ethernet demarcation Infinera’s Range of Packet- transparent Layer 1 services. In addition, over the Ethernet network (Figure 31). as described in section 5.4. unit [EDU] or network interface device [NID]) FIGURE 32: Infinera’s Network Optical Transport Switches an Infinera packet-optical network can deployed at the customer location (Figure 32). Interface Device (NID) (EMXP) in the XTM Series, seamlessly interoperate with its transport These units enable a clear separation between a Chassis-based Unit and a network over MPLS-TP tunnels, Ethernet the user and provider Ethernet networks. 1RU Access Unit Protection 2.6 THE MAIN ELEMENTS OF AN SVLANs or OTN switches and any To maintain SDH/SONET protection, INFINERA PACKET-OPTICAL combination of these. the existing Automatically Switched NETWORK Optical Network (ASON) and subnetwork Having looked at the principles of packet- connection protection (SNCP) protection optical networking, it is time to discuss
34 INFINERA INFINERA 35 PACKET-OPTICAL NETWORKING
capabilities across the metro aggregation include Layer 1 transponders and management. The real benefits of packet- A further development of the EMXP and core networks. muxponders, EMXPs, ROADMs and other optical networking can only be realized concept is the Packet Transport Fabric optical network elements. with a truly integrated Layer 1 and Layer 2 (PT-Fabric), where a high-capacity EMXP The metro core network usually connects to transport platform and a unified Layer 1 and is interconnected to external 10 Gb/s and data centers, and also to IP core and mobile Tributary Modules Switch Modules Line Modules Layer 2 management system. 100 Gb/s input/output (I/O) client units OTN core networks, typically implemented by OTN 12 2.6.4 The DTN-X Packet-Switching via a separate frontplane bus. The optical a set of IP routers using full IP/MPLS for OTN A network for Carrier Ethernet services frontplane takes vertical-cavity surface- Module (PXM) OTN traffic engineering purposes. Normally OTN may be implemented as a separate emitting laser (VCSEL) technology used in Client ODUk ODUk <-> ODUJ ODUj ODUj N x ODUj Super-Channel operators prefer to have a clear demarcation (e.g. ODU0) (e.g. 5x ODU4) Layer 1 optical network with Layer 2 supercomputing, and brings it to optical The metro/regional packet-optical network (e.g. 10GbE) (e.g. ODU2) (e.g. ODU2 <-> ODU0) (e.g. ODU0) (e.g. ODU0) transport equipment for the first time. point toward the metro network at the described above can be further augmented, Ethernet switches attached externally edge of the IP/mobile core network, often both in the metro core and regarding over standard interfaces, as depicted in This frontplane enables the PT-Fabric to be referred to as a provider edge router. The integration with long-haul optical networks, OTN OTN OTN WDM Figure 35. Although conceptually simple, deployed in any combination of available slots Mapping Multiplexing Switching Super-Channels routers of IP/mobile core networks have by Infinera’s DTN-X Family of converged such a configuration results in a complex in single or multiple XTM Series chassis and broad functionality and very high capacity; digital photonic platforms. The DTN-X Family hierarchy of management systems. These without any backplane capacity limitations. is a terabit-capable super-channel system Block Diagram Showing the Main Functions of the Infinera DTN-X Family Multiple PT-Fabrics can be deployed in a they are therefore normally interconnected FIGURE 34: systems must be carefully integrated in leveraging large-scale photonic integrated single chassis, giving a total of 4 terabits of via a strict Layer 1 WDM network, since the circuit technology, integrating OTN switching switching within one XTM Series chassis. main objective is to provide “fat pipes” and WDM transport. With the inclusion of without any need for Ethernet aggregation. the PXM, the DTN-X family is transformed for long-haul purposes or in a densely 2.6.5 The Multi-layer Service FIGURE 35: Ethernet Services into a fully packet-aware platform where populated area, and where the packet- Management System Provided by Separate Ethernet Switches Attached to an Optical The metro aggregation and metro core packet flows can be aggregated and switched mode traffic now can be transported by The chosen network architecture has 2.6.3 Optical Add-Drop Multiplexers Network Result in a Complex networks are interconnected via the XTM on individual ODUs to facilitate traffic individual flows fully managed according a profound influence on the degree of Hierarchy of Management Systems Series packet-optical platform, which can and Other Optical Elements engineering (Figure 34). to OTN standards. Furthermore, the PXM operational simplicity that it is possible The Layer 2-specific elements of metro provide switching at OSI Layers 1 and is certified for and can terminate CE 2.0 to achieve when it comes to network Element Management aggregation and metro core networks System for the 2. In the metro core, Ethernet SVLANs services such as E-Line and E-LAN, making packet layer and MPLS-TP enable detailed handling use Layer 1 optical WDM channels for The PXM maps the Ethernet VLANs and it possible to set up Ethernet services Service Management of Layer 2 services, while WDM keeps the transport of Ethernet frames between MPLS services of the packet-optical network starting in an NID in the access network and System legacy transport services at Layer 1. A network nodes, as described in section to the Layer 1 OTN transport services. This terminating via a DTN-X PXM in the metro 2.2.3. All the above Layer 2-specific network mapping is especially cost-saving when Layer 2 (Packet) ROADM-enabled Layer 1 provides flexible core or long-haul network. More about how Integration wavelength switching capabilities, while a elements interwork seamlessly with the an OTN system is already in place, e.g. the PXM maps Ethernet traffic to OTN can Layer 1 (Optical) unified control plane provides management flexible optical networking elements at be found in section 2.2.5. Layer 1 when present in a truly integrated
packet-optical platform, such as Infinera’s Element Management System for the XTM Series. An XTM Series node may optical layer
36 INFINERA Footnote citations on page 111 INFINERA 37 PACKET-OPTICAL NETWORKING