Evolution of Ethernet Standards in IEEE 802.3 Working Group

| Marek Hajduczenia, Steven B. Carlson, Dan Dove, Mark Laubach, David Law, George A. Zimmerman, David Law

Abstract:

Ethernet is constantly evolving, adapting to the needs of the networking world and addressing the requirements of both operators and end-users, while making sure that the resulting technology is cost-efficient, reliable, and operates in a plug and play manner. The IEEE 802.3 Working Group has been working for the last 30+ years, pushing the boundaries on the speed and capacity of wireline Ethernet links, migrating from shared medium CSMA/CD systems to switched point-to-point Ethernet, and then introducing multilane technology and point-to-point emulation over shared media of passive optical networks (PON). In this paper, we look at the latest projects adding new features and capabilities to the family of wired Ethernet standards that enable the exponential growth of the Ethernet ecosystem and are driven by technical maturity, cost-effectiveness, and broad market support.

Index Terms: Ethernet, 802.3 Working Group, 2.5/5G, 25G, 40G, 100G, TimeSync, EPON, EPoC, RTPGE, YANG, backplane.

 

I.       Introduction

The total amount of data created or replicated on the planet in 2010 exceeded 1 zettabyte (1 zettabyte is 1021 bytes)–that is 143 GB for each of the 7 billion people on the planet [1]. This volume of information requires high-speed links between server farms, cloud storage, and end-users to make sure that it can be processed in a timely and reliable fashion. The relentless growth of the number of permanent or nomadic end-stations connected to the network (e.g., computer terminals, mobile devices, automated devices generating machine-to-machine traffic) has led to explosive growth in the volume of information exchanged at all levels of the networking infrastructure. The popularity of Ethernet and its widespread use in access, aggregation, transport, core networks, and data centers, combined with the unprecedented demand for advanced data connectivity services, fuel the development of new Ethernet standards, providing higher-speed links to address market demand.

Ethernet is also venturing into brand-new application areas and adding support for synchronization protocols. Potentially, Ethernet could become the de facto standard for in-vehicle data networks, providing a common transport platform for control and multimedia applications.

This paper examines the evolution of Ethernet standards taking place in the IEEE 802.3 Working Group. There are a number of exciting new projects, pushing the boundaries of Ethernet into new application areas and markets.

 

II.      Evolution of Ethernet Standards

The 802.3-2012-IEEE Standard for Ethernet was first published in 1985, specifying a half-duplex carrier sense multiple access with collision detection (CSMA/CD) media access control (MAC) protocol operating at 10 Mb/s, and a medium attachment unit (MAU) for operation on a coaxial cable medium supporting a bus topology between the attached end stations.

Amendments to IEEE 802.3 then added specifications for, among other items, a repeater to extend the topologies supported, MAUs for operation over fiber optic cabling, a MAU for operation over twisted pair cabling, 10BASE-T, and layer management. In 1995, Amendment IEEE 802.3 was published adding operation at 100 Mb/s (fast Ethernet). This included a number of physical layer entity (PHY) specifications for operation over fiber optic and twisted pair cabling (100BASE-TX).

Amendment IEEE 802.3x, published in 1997, added full-duplex operation to the MAC. A flow control protocol was also added to take advantage of full-duplex-capable media such as twisted pair and fiber, for which PHYs had already specified for in IEEE 802.3, as well as support switching, which was becoming more cost-effective due to increased device integration.

In 1998, Amendment IEEE 802.3z was published to add operation at 1,000 Mb/s (Gigabit Ethernet). In 1999, Amendment IEEE 802.3ab was published to add 1000BASE-T PHY specifications to support 1,000 Mb/s operation over twisted pair cabling.

Amendment IEEE 802.3ad (link aggregation) was published in 2000, adding the ability to aggregate multiple full-duplex point-to-point links into a single logical link from the perspective of the MAC client. Since link aggregation has applications beyond Ethernet, as well as architectural positioning, it was subsequently moved to the IEEE 802.1 Working Group in 2008 and is now titled IEEE 802.1AX Link Aggregation.

In 2002, Amendment IEEE 802.3ae was published to add operation at 10 Gb/s (10 Gigabit Ethernet). In 2006, Amendment IEEE 802.3an was published adding 10GBASE-T PHY specifications to support 10 Gb/s operation over twisted pair cabling. It was followed in 2010 by Amendment IEEE 802.3ba, which added operation at 40 Gb/s and 100 Gb/s (40 Gigabit Ethernet and 100 Gigabit Ethernet). The development of 40 Gb/s and 100 Gb/s Ethernet was done in close cooperation with ITU-T Study Group 15-Networks, Technologies and Infrastructures for Transport, Access, and Home to ensure transparent connectivity into the optical transport network (OTN); operation at 10 Gb/s, 40 Gb/s, and 100 Gb/s only supports full-duplex operation.

Amendment IEEE 802.3ah was published in 2004 to add support for subscriber access network Ethernet (Ethernet in the first mile, or EFM for short). This amendment added a number of fiber optic and voice grade copper PHYs and specified a fiber optic point-to-multipoint network topology using passive optical splitters, which is known as Ethernet passive optical network (EPON).

In 2007, Amendment IEEE 802.3ap was the first to add support for backplane Ethernet.

A summary of the speed and distance for various MAUs and PHYs supported by the approved IEEE 802.3 standard (at the time this paper was written) and amendments is shown in Fig. 1.

Other additions include IEEE 802.3af DTE Power via MDI, which was published in 2003, and is also known as power over Ethernet, which enables power to be supplied on the same cabling as the data transmission. IEEE 802.3at was published in 2009 and enhanced the maximum power available and the classification mechanism.

In addition, the 2010 amendment IEEE 802.3az added support for energy-efficient Ethernet (EEE) to the 100BASE-T, 1000BASE-T, and 10GBASE-T PHYs, among others. This not only reduces the power consumption of the PHYs, but also specifies signaling that can enable the reduction of the power consumption of the attached device.

Figure 1: Speed and reach for various IEEE Std 802.3 MAUs and PHYs

Figure 1: Speed and reach for various IEEE Std 802.3 MAUs and PHYs

 

A.   High-speed Copper P2P Links and Backplane Technologies

In 2007, the family of PHYs for Ethernet operation over electrical backplanes for Gigabit Ethernet and 10 Gigabit Ethernet was first introduced. Two PHYs were introduced for 10 Gigabit Ethernet: 10GBASE-KX4 and 10GBASE-KR. The 10GBASE-KX4 PHY is a full-duplex solution employing four data lanes in each direction, where each lane operates at 3.125 Gb/s and employs 8B/10B line encoding to support the effective data rate of 10 Gb/s. The 10GBASE-KR PHY is a serial lane solution operating at 10.3125 Gb/s and employing 64B/66B encoding to support the effective data rate of 10 Gb/s.

These two PHYs laid the groundwork for the 40 Gigabit Ethernet backplane PHY, which was developed during the IEEE P802.3ba project. Using the four lane approach of 10GBASE-KX4 and the serial 10 Gb/s electrical signaling developed for 10GBASE-KR, the 40GBASE-KR4 PHY supports 40 Gigabit Ethernet operation across an electrical backplane. At the time of the IEEE P802.3ba project no 25 Gb/s per lane electrical signaling solution was available, therefore no 100 Gigabit Ethernet backplane PHY was developed.

The call-for-interest (CFI) to develop the operation of Ethernet at 100 Gb/s across an electrical backplane, as well as across twin-axial cables, took place in November 2010. Fueled by the SFP+ form factor supporting 10 Gigabit Ethernet, QSFP supporting 40 Gigabit Ethernet, or the CXP or CFP supporting 100 Gigabit Ethernet, potential front-panel capacities ranging anywhere from 480 Gb/s to 3.2 Tb/s were observed. These front-panel capacities could create backplane requirements ranging anywhere from 3.2 Tb/s to 44.8 Tb/s depending on the specific system configuration. A comparison was made between the then-existing 10 Gb/s signaling technologies against a potential 25 Gb/s signaling to understand the impact on the total number of copper of differential pairs needed to support various backplane capacities. Fig. 2 illustrates the impact of 10 Gb/s versus 25 Gb/s on the total number of differential pairs needed when supporting various backplane capacities for various switch fabric configurations. Note that as the capacity requirement of the backplane increases, the ability to support an actual total capacity of the backplane with 10 Gb/s per lane signaling becomes questionable [9].

 

The challenge with electrical backplanes as compared to copper cabling is that they are essentially custom-designed. There are a multitude of factors that influence the electrical performance of the backplane channel: FR4 board materials, trace geometries, surface roughness of the copper traces, and the actual system configuration among other characteristics. This is further complicated by the cost sensitivity of channels, where material costs alone could increase the cost of a backplane by 500% depending on the materials compared [10].

The large variation in electrical performance and sensitivity to cost resulted in the development of 100 Gigabit Ethernet backplane objectives targeting different performance/cost targets:

  • Define a four lane PHY for operation over a printed circuit board backplane with a total channel insertion loss of <= 35 dB at 12.9 GHz.
  • Define a four lane PHY for operation over a printed circuit board backplane with a total channel insertion loss of <= 33 dB at 7.0 GHz.

IEEE Standard for Ethernet 802.3bj-2014 specifies an NRZ (nonreturn to zero)-based solution for the 35 dB @ 12.9 GHz objective, and a PAM-4 based solution for the 33 dB at 7.0 GHz objective. In addition, the same amendment added a 100 Gigabit Ethernet x4 twin-ax cable solution, defining a four-lane 100 Gb/s PHY for operation over links consistent with copper twin-axial cables with lengths up to at least 5 m. An NRZ-based solution was selected to address this objective. These new PHYs included support for an optional energy-efficient Ethernet mode.

The four lane 100 Gb/s connections defined in IEEE 802.3bj-2014 laid the groundwork for 25 Gb/s signaling over a single lane.  High-performance server interconnects rapidly seized on the technology, and in July 2014 a study group was formed to standardize a 25 Gb/s single-lane Ethernet. IEEE Standard for Ethernet 802.3by-2016 was built on the basis of 802.3bj-2014 and specified 25 Gb/s solutions for use on copper backplanes, copper twin-axial cables with lengths of 3m and 5m, and multimode fiber solutions for 100m. Building on existing technology, IEEE 802.3by-2016 went from CFI to ratification in 23 months.

Fig. 2. Impact of signaling speed on switch capacities.

Fig. 2. Impact of signaling speed on switch capacities.

 

B. Optical P2MP Links and Evolution of EPON

EPON is a relatively new addition to the family of Ethernet standards, with the first standard for this technology (1G-EPON, operating at the symmetric data rate of 1 Gb/s) published in 2004. In 2009, a higher-speed version of EPON was standardized, supporting the symmetric data rate of 10 Gb/s as well as an asymmetric data rate of 10 Gbit downstream (towards the customer) and 1 Gb/s upstream (from the customer). Supporting the nominal distances of 20 km (or more) and the nominal split of 1:32 (or more) with three available power budget classes, EPON is used in a variety of deployment scenarios, some of which are shown in Fig. 3. More details about EPON technology can be found in [2] and [3], including a definition of the individual power budget classes.

Fig. 3. Examples of various EPON deployment scenarios.

Fig. 3. Examples of various EPON deployment scenarios.

The observed ~50% annual growth in volume of Internet traffic in residential applications is driving the migration from legacy to fiber-based access technologies. For the residential subscribers served by EPON, the speed of residential wired or wireless local area networks (LANs) becomes the primary gating factor for the bandwidth demand. While being predominantly in the range between 100 Mb/s and 1 Gb/s today, the interface speeds of customer equipment (PCs, laptops, set-top boxes, TVs, security cameras, personal storage farms, etc.) are expected to increase by 2.5–5.0 Gb/s with the advent of IEEE 802.3bz 2.5/5-GBASE-T interfaces.

While unified in the common trend to support more subscribers with higher data rates, the residential access, business access, and mobile backhaul markets have different bandwidth targets and technical performance requirements. Not only are the technical requirements different in all of these markets, but the cost-to-performance objectives are also different. The IEEE P802.3ca Next Generation EPON Task Force was created to address these diverse requirements:

  • A multiwavelength (per-direction) EPON PHY (i.e., hybrid PON) with an aggregate downstream capacity of at least 40 Gb/s (40G-EPON), with an evolutionary path to 100 Gb/s (100G-EPON).
  • A single wavelength (per direction) EPON PHY (i.e., TDM-PON) that supports symmetric downstream and upstream line rates of at least 25 Gb/s (25G-EPON) or 25 Gb/s downstream and 10 Gb/s upstream (25/10G-EPON).

Coexistence with 10G-EPON on the same ODN, support for multiple generations of equipment, as well as a flexible and extensible standard definitions are examples of other critical requirements for this new technology. The task force is still in the early stages of technical development, focusing on a power split multirate P2MP architecture, as shown in Fig. 4, where a single optical line terminal (OLT) capable of multirate operation supports optical network units (ONUs) with different data-rate capabilities depending on the number of supported wavelength channels.

Fig. 4. NG-EPON OLT supporting multiple NG-EPON ONUs.

Fig. 4. NG-EPON OLT supporting multiple NG-EPON ONUs.

 

Given the interest in the development of a single extensible standard to support multiple data rates and generations of NG-EPON devices, it is likely that the multipoint control protocol (MPCP), used extensively in 1G-EPON and 10G-EPON for the purpose of station discovery and bandwidth allocation, will be extended in NG-EPON and will also perform device capability discovery and wavelength channel negotiation.

More details about the progress of the IEEE P802.3bc Task Force can be found at: http://www.ieee802.org/3/ca/index.html.

 

C.   EPON Protocol over Coax: Bringing the Copper and Optical Worlds Together

Deployment of gigabit-capable EPON (based on 1G-EPON and 10G-EPON) services by cable operators in both China and the United States has been increasing over the past several years. The market drivers in both markets are slightly different. Hybrid fiber-coaxial (HFC) deployments as well as DOCSIS [4] are not as widely deployed in China as in other parts of the world. Chinese cable operators are looking at the opportunity of transparently extending EPON services over legacy coaxial cabling in multi-tenant/dwelling units (MxU) and businesses. In North America, high-speed data (HSD) residential services are currently provided using DOCSIS technology. However, for competitive multigigabit business class services, cable operators are increasingly deploying EPON to capture market share leveraging metro Ethernet forum (MEF) [5] service performance and competitive service level agreements (SLAs), all managed by DOCSIS provisioning of EPON (DPoE™ [4]) technology, which was developed jointly by operators, vendors, and the company CableLabs.

Another trend in the worldwide cable network industry is the step-wise migration from backend legacy MPEG-2 transport to MPEG-4 video distribution via IP over Ethernet. In the future, a large Ethernet-based gigabit pipe to the home and business will be fundamental for cost-effective growth and evolution. Fig. 4 shows examples of some of the target applications of the mix of EPON and EPoC technologies, leveraging the fiber-deep access architecture of current networks and also reusing existing coaxial distribution infrastructure to the greatest extent possible.

Both Chinese and North American operators share the desire for the simplicity of Ethernet at gigabit speeds, and collectively asked the IEEE 802.3 Ethernet Working Group [6] to create a standards effort for extending the operation of EPON protocols over coaxial distribution networks, a project that was called EPON Protocol over Coax, or EPoC for short. There are many opportunities where EPON has been deployed adjacent to or alongside existing coaxial networks, and some customers are more opportunistically reached by simply extending EPON over coax. The key to this transparent extension is unified management, service, and quality of service (QoS).

Fig. 5. EPoC applications for extending EPON services over coax.

Fig. 5. EPoC applications for extending EPON services over coax.

 

For EPON, IEEE 802.3 standards define the MAC and PHY sublayers for a service provider OLT and a subscriber ONU. The fiber optical interconnecting media uses two wavelengths for full-duplex operation, one for continuous downstream channel operation and another for upstream burst mode operation. The OLT MAC controls time-division sharing of the upstream channel for all ONUs.

Similarly, the EPoC architecture consists of a service provider coax line terminal (CLT) and a subscriber coax network unit (CNU). The EPoC CLT and CNU MAC sublayers will be substantially similar (if not the same) to the layers found in the OLT and ONU, respectively. A new PHY will be specified for operation over the coaxial distribution network (CxDN) media. Downstream and upstream communication channels will use the radio frequency (RF) spectrum as assigned by a cable operator for their coax network.

Two system models are supported by EPoC, as shown in Fig. 5. The first is a CLT with one more CNU interconnected by a coaxial distribution network. The second is enabled by the future EPoC standard, but is outside the scope of the IEEE 802.3 Working Group. That is a traditional EPON with an OLT and multiple ONU devices together with one or more optical-to-coax media converter devices that attach between the PON and a CxDN using EPoC, permitting CNUs to appear as ONUs to the OLT. The industry will likely create new products using the second model.

For more information about the task force, please see http://www.ieee802.org/3/bn/index.html.

Fig. 6. EPoC standard and EPoC-enabled system models.

Fig. 6. EPoC standard and EPoC-enabled system models.

 

D.   Timing and Synchronization in Ethernet (TimeSync)

Support for synchronization in Ethernet rapidly becomes a critical feature, especially due to requirements of digital content distribution, video and audio systems with remote streaming, or even mobile backhauling. All of these application areas require not only delay-guaranteed, engineered, and strictly controlled links (in terms of QoS, bandwidth, and jitter), but also the ability to synchronize with a common reference clock to assure proper operation of specific features of the given application. To address these requirements, the IEEE P802.3bf Task Force was created in 2009 to develop a method for “an accurate indication of the transmission and reception initiation times of certain packets, as required to support IEEE P802.1AS” [7].

The resulting architecture is presented in Fig. 6. This project added the following new features to the IEEE 802.3 architecture:

  • Rx SFD detect and Tx SFD detect functions, responsible for detecting the reception and transmission of an Ethernet frame, respectively, and relaying this information to upper layers (TimeSync client) via the TSSI (TimeSync service interface).
  • Set of managed objects and registers, providing the TimeSync client with the ability to read ingress and egress latency information characteristic of the given PHY. This provides the TimeSync client with the ability to perform necessary synchronization calculations relative to the reference plane located at the bottom of the 802.3 stack at the media dependent interface (MDI).
Fig. 7. Relationship between IEEE 802.3bf functions, TSSI, and remaining IEEE 802.3 layers. All clause numbers are relative to IEEE 802.3.

Fig. 7. Relationship between IEEE 802.3bf functions, TSSI, and remaining IEEE 802.3 layers. All clause numbers are relative to IEEE 802.3.

 

The 802.3bf-2011-IEEE Standard for Information Technology architecture was designed to provide direct support for the 802.1AS-Timing and Synchronization TimeSync client operating on top of IEEE 802.3 PHYs. However, it was quickly discovered that potential applications of the newly-specified TSSI could also cover other synchronization protocols, e.g. IEEE 588v2 and other proprietary use cases, which can benefit from information about transmit and receive path latencies as well as identification of the frame transition event through the RS sublayer.

The potential use of IEEE 802.3bf to support IEEE 1588v2 (IEEE 1588 Standard for A Precision Clock Synchronization Protocol for Networked Measurement and Control Systems) resolves one of the long-standing problems of this specific synchronization protocol, namely the lack of a standardized way to retrieve correlated information between the frame transmission time and synchronized time. Various proprietary mechanisms have been developed over the course of the last few years, some of them quite similar to the solution proposed in IEEE 802.3bf. It is expected that the TSSI will become a de facto standard for future implementations of the IEEE1588v2 protocol operating on top of Ethernet PHYs.

IEEE 802.3bf is now part of 802.3-2015 – IEEE Standard for Ethernet (see Clause 90).

 

E. Ethernet on Twisted Pairs: BASE-T Differentiates

Following the standardization of IEEE 802.3an-2006 defining 10GBASE-T, and IEEE 802.3ba-2010 defining 40 Gb/s and 100 Gb/s Ethernet, work began in the industry to standardize the next higher speeds of twisted-pair Ethernet. Following a successful CFI in July 2012 and subsequent study group, the IEEE P802.3bq Task Force began work on 40GBASE-T. With the advent of 25 Gb/s Ethernet, this work expanded to include 25GBASE-T. Focusing on data center middle-of-row and end-of-row architectures (see Fig. 6), 802.3bq-2016-IEEE Approved Draft Standard for Ethernet maintains the backwards compatibility of the standard Ethernet RJ-45 connector and the support for the auto-negotiation function, which has made BASE-T Ethernet successful. This standard adds two new BASE-T PHYs, one for 25 Gb/s and another for 40 Gb/s operation over Category 8 (ISO/IEC Class 1 or Class 2 channels) cabling at a distance of up to 30 meters. The technology base of 10GBASE-T proved useful to this project, and in addition to the increased bandwidth of the cabling, the standard made minor improvements to the error correction coding and startup parameter exchanges in 10GBASE-T. 10GBASE-T, 25GBASE-T, and 40GBASE-T; both support Clause 28-Auto-Negotiation and Energy Efficient Ethernet.

Fig. 8. ToR, MoR, and EoR interconnection options.

Fig. 8. ToR, MoR, and EoR interconnection options.

The demand for higher-speed backhaul for IEEE 802.11 wireless LAN access points and the need to use in-line power drove the IEEE 802.3 WG to standardize two new lower speeds for BASE-T at 2.5 Gb/s and 5 Gb/s. Two of the primary requirements included operation on the large installed base of Category 5e and Category 6 cabling as well as operation above 1 Gb/s and below 10 Gb/s. The work of the IEEE P802.3bz Task Force showed a lot of synergies with the IEEE 802.3bq-2016 standard, and its resulting technical solution is largely based on 10GBASE-T technology with some differences in the modulation and coding structure. The (at the time of this writing) new IEEE P802.3bz amendment draft defines MAC parameters for two new speeds (2.5 Gb/s and 5 Gb/s), both using an extension of the 10 Gb/s MAC interface (XGMII). The draft also specifies two PHYs, operating at speeds of 2.5 Gb/s and 5 Gb/s over Category 5e, Category 6, or better cabling at distances of up to 100 meters. The new 2.5GBASE-T and 5GBASE-T PHYs offer a bridge to higher speeds for existing 100-meter BASE-T networks and intermediate speed operation on newer networks. At the time of this writing, IEEE P802.3bz 2.5G/5GBASE-T Draft 3.1 (see http://www.ieee802.org/3/bz for more details) is in the final stages of sponsor ballot recirculation, and is expected to be approved in September 2016.

 

F.    Ethernet in New Markets: Applications in the Automotive Industry

The IEEE 802.3 family of standards provides a wide variety of solutions for data networks with many different operating speeds over copper wire, electrical backplanes, and various optical media. Recently, the global automotive industry has decided to deploy Gigabit Ethernet as a network backbone in automobiles and light trucks by the year 2020 [8]. 1000BASE-T as defined in IEEE 802.3 uses four twisted copper wires pairs. While this is not an issue for structured wiring plants, it results in a cable that is too heavy, costly, and cumbersome for vehicular use. (See slide 15 in [8] for an example of a typical passenger vehicle harness.)

The IEEE P802.3bp 1000BASE-T1 Task Force has completed the development of a robust 1 Gb/s copper PHY for this new market area. Estimates place the number of Ethernet ports in cars at around 300 million ports per year by 2019. The 1,000BASE-T1 PHY will allow for smaller and lighter cabling and the use of a network backbone architecture that will make the in-car wiring harness easier to manufacture and lower in cost. In fact, the wiring harness in a car is the third most expensive component in the car behind the engine and chassis, and is also the third heaviest.

In 2012, modern cars had between 40 and 60 microcontrollers, while high-end cars had over 120. Future sophisticated camera and control systems, vehicle safety devices (automatic braking, collision avoidance, etc.), infotainment, and GPS systems will create large traffic volume for the in-car network that the previous automotive networks could not handle.

The number of microcontrollers or electronic control units (ECUs) in cars is expected to rise dramatically over the next decade as these functions become standard features in new cars. Automotive networks will also leverage other IEEE 802.3 technologies such as IEEE 802.3az and IEEE 802.3bf (discussed in the previous section) to provide an optimized network solution.

802.3bp-2016-IEEE Approved Draft Standard for Ethernet was approved in June 2016 and is being designed into the cars that will appear in the early 2020s.

But there is much more to the story. The IEEE P802.3bp project generated so much interest in the Ethernet community that the IEEE P802.3bu 1-Pair Power over Data Lines (PoDL) project was initiated to provide DC power over the same single twisted wire pair used for IEEE 802.3bp-2016. IEEE P802.3bu is expected to finish in early 2017. A proprietary 100 Mb/s Ethernet solution, the OPEN Alliance BroadR-Reach compliant Ethernet PHY, was brought into the IEEE 802.3 Ethernet Working Group and standardized as 802.3bw-2015-IEEE Standard for Ethernet Amendment 1. In addition, work on a 1 Gb/s solution on plastic optical fiber (POF) is being done in the IEEE P802.3bv Gigabit Ethernet Over Plastic Optical Fiber Task Force, with an expected completion date in 2017.

 

G.   Management for Ethernet Networks

Ethernet as defined by the IEEE 802.3 Working Group continues to evolve by adding support for higher data rates, new media types, and new features. Ethernet will expand into new markets and address new application areas, as discussed in the previous sections. However, this evolution may require changes in the managed objects stored in the management information base (MIB), allowing management systems to take full advantage of the newly-added Ethernet features.

In order to provide a consistent, up-to-date version of MIB definitions and eliminate dependence on external MIB definitions produced outside of the IEEE 802.3 Working Group, a project was started at the end of 2008 (IEEE P802.3be) targeting organization, updates, and the consolidation of managed object definitions provided in IEEE 802.3-2008, including the logical link discovery protocol Ethernet extensions provided in IEEE 802.1AB-2009, Annex F. In addition, the initial version of this standard incorporated and updated the MIB module definitions formerly defined in a series of RFC documents, namely RFC 2108, RFC 3621, RFC 3635, RFC 3637, RFC 4836, RFC 4837, RFC 4878, and RFC 5066. The final version of IEEE 802.3.1 was published in July 2011, containing both the definitions of individual MIBs and associated descriptions as well as the machine-readable MIB files, which are available from the website of the IEEE P802.3be project. The published standard was then amended to account for amendments to IEEE 802.3-2008, including IEEE 802.3at, IEEE 802.3av, IEEE 802.3az, IEEE 802.3ba, IEEE 802.3bc, IEEE 802.3bd, IEEE 802.3bf, and IEEE 802.3bg. The resulting 802.3.1-2011-IEEE Standard for Management Information Base (MIB) Definitions for Ethernet was published in June 2013, comprising the latest version of Ethernet MIB.

Apart from maintaining MIB for legacy SNMP-based management systems, the IEEE 802.3 Working Group is also actively pursuing more modern management schemes for Ethernet devices. YANG (yet another next generation data model, see RFC 6020) is quickly becoming the de facto data modeling language for next generation network management systems, replacing the legacy management (MIB)/simple network management protocol (SNMP)-based tools. YANG is a data modeling language, which replaces the rigid structure of MIB with a very flexible and extensible way to describe different data types, aggregating them into different object types. It is used to express, for example, interfaces, devices, network topology, or even protocol models, and builds on existing models to create more complex data structures. YANG data models describe configuration, monitoring, administration, and notification capabilities in a device-independent but end-to-end network service-oriented manner, providing network management in a simple, human-readable language syntax.

When combined with a reliable transport protocol (e.g., NETCONF), YANG provides substantial advantages to operators, simplifying end-to-end network deployment, and providing vendor-independent service modeling across different hardware platforms.

The development of YANG data models has seen incredible growth in many industry organizations such as the IETF, Metro Ethernet Forum, and the IEEE 802.1 Working Group. Within the IEE 802.3 Working Group, the work on YANG models for Ethernet will be undertaken by the recently-formed YANG Data Model(s) Study Group. (See http://www.ieee802.org/3/ce/index.html for details.)

 

III.    Summary

The work within the IEEE 802.3 Working Group is far from done, with the next generation of high-speed 40/100/200/400G links aiming for broader market adoption through increasing the cost-effectiveness of solutions while decreasing the power consumption and complexity of compatible products. This work also focuses on lower speeds. The 10 Mb/s Extended Reach Single Twisted Pair Ethernet PHY (http://www.ieee802.org/3/cfi/request_0716_1.html) project, aims to address existing market demand for a unified lower speed and a longer-reach PHY for automation purposes. The IEEE 802.3 Working Group is thus looking for ways to expand Ethernet market coverage and to support higher data rates while also providing coverage for emerging markets such as the automotive industry.

It can be expected that innovation in the area of wired Ethernet will continue in the years to come, bringing the same highly reliable and well-understood networking philosophy to new markets, enabling new applications, and making networking in general more ubiquitous.

 

References


Marek Hajduczenia

marek.hajduczenia@charter.com

Marek Hajduczenia is the Network Architect and Principal Engineer at Charter Communication and focuses on R&D for EPONs. He is involved in the IEEE 802.3 WG, focusing on technical and editorial contributions to numerous projects, including IEEE Std 802.3av-2009 (10G-EPON), IEEE 802.3bf-2011 (TimeSync for Ethernet), IEEE P802.3bk (Extended EPON), IEEE P802.3bn (EPoC) and many others. He supports IEEE P1904.1 SIEPON with technical contributions and acts as Chief Editor. He also participates in CableLabs in development of the DPoE™ series of specifications. He received his Ph.D. degree in the area of electronics and telecommunication from the University of Coimbra, Portugal. Currently, he holds more than 35 international and European patents.


Steven B. CarlsonSteven B. Carlson 
scarlson@ieee.org
Steven B. Carlson is the President of High Speed Design, Inc., a Portland, Oregon-based consulting company. Mr. Carlson has over 40 years’ experience in embedded control systems and networking for the entertainment and energy management industries. He currently serves as the Chair of the IEEE P802.3bp 1000BASE-T1 Task Force, the P802.3bw 100BASE-T1 Task Force, and is the Executive Secretary of the IEEE 802.3 Ethernet Working Group. Mr. Carlson previously served as the Chair of IEEE 802.3af-2003 DTE Power via MDI project, usually referred to as “Power over Ethernet,” the IEEE 802.3bf – 2011 Time Sync Task Force and was a founder of the Entertainment Services and Technology Association’s Technical Standards Program, ANSI E1-Entertainment Technology.


Dan Dove

dan.dove@dovenetworking.com

Dan Dove is Chair of the IEEE 802.3bm Task Force, which is working to define the next generation of 40G and 100G Ethernet optical technologies. He has been working at Applied Micro for a year, helping them to establish a technology strategy that will position them for success in data center networking. In addition to his responsibility to identify new technology opportunities, he represents them in key networking consortiums and standards bodies including IEEE 802.3 and OIF, and as a member of the Board for the Ethernet Alliance. Prior to his employment at Applied Micro, he was principal engineer for Physical Layer Technologies at Hewlett Packard’s Networking Business unit. He worked at HP for 31 years, spanning a career that started in production assembly and worked through a technical ladder to his ultimate role as a master-level engineer. He has been working in the IEEE 802.3 Working Group since 1988, participated in many projects, and led the 802.3ak 10GBASE-CX4 project to successful completion. He holds 21 patents including the broadly used Auto-MDIX patent used for twisted pair physical layers.


Mark_Laubach_1_800Mark Laubach

mark.laubach@broadcom.com

Mark Laubach is the Chair of the IEEE P802.3bn Task Force on EPON Protocol over Coax standards effort.  At Broadcom Ltd., his work covers broadband management, architecture, standards efforts for EPoC systems and Next Generation EPON systems.  Mark has more than 20 years of experience spanning broadband, gigabit, and cable access network technologies including being CEO and President at Broadband Physics, VP and CTO at Com21, and a senior engineer at HP Labs.  He is an inventor on several U.S. patents on cable modem and broadband technology. His past activities in standards include IEEE, IETF, SCTE, and the ATM Forum.  He is an IEEE Fellow and an SCTE Senior Member. He is the principle co-author of a cable modem technology book “Breaking the Access Barrier: Delivering Internet Connections Over Cable” (John Wiley Press, 2000). He holds a BSEE and MSCS from the University of Delaware under Professor David J. Farber.


 

Photo of David LawDavid Law

dlaw@hpe.com

David Law is a Distinguished Technologist at Hewlett Packard Enterprise and has worked on the specification and development of Ethernet products since 1989. Throughout that time he has been a member of the IEEE 802.3 Ethernet Working Group, where he has held a number of leadership positions, including Chair of IEEE 802.3 since 2008 and Vice-Chair between 1996 and 2008. He received the IEEE-SA Standards Medallion, the IEEE-SA Standards Board Distinguished Service Award, the IEEE-SA International Award, the 2016 IEEE Computer Society Karlsson Award, and the 2017 IEEE Charles Proteus Steinmetz Award. He has a BEng (hons) in Electrical and Electronic Engineering from Strathclyde University in Glasgow, Scotland.


ZimmermanGeorge A. Zimmerman

george@cmephyconsulting.com

George Zimmerman is an independent consultant, specializing in physical layer communications technology. Dr. Zimmerman’s technical interests focus on communications and signal processing near the fundamental limits and energy efficiency at the PHY and system levels. He holds a Ph.D. in Electrical Engineering from Caltech, and an undergraduate degree from Stanford University. He is currently active in the IEEE 802.3 working group, and is the editor for IEEE 802.3bq-2016 25/40GBASE-T and the IEEE P802.3bz 2.5G/5GBASE-T projects, as well as active in the IEEE P8023.bt 4 Pair Power over Ethernet Task Force and various other 802.3 projects. He serves on the board of the NBASE-T Alliance, as well as Technical Committee Chair for the Ethernet Alliance.  He has been a defining force in the development of multiple Ethernet technologies, including 10GBASE-T, Energy Efficient Ethernet, as well as various DSL (now EFM) technologies. He founded Solarflare Communications as CTO of its PHY group, and holds more than a dozen patents related to efficient implementations of high-speed Ethernet technology.  As a consultant, his clients include networking systems, cabling infrastructure, and physical-layer silicon companies.