Telecom industry preps for 100G DWDM optical transport


Telecom industry preps for 100G DWDM optical transport

Editor's Note: Telecom service providers have made it clear they need 100G transport technology in their long-haul networks in order to handle increased customer traffic and bandwidth use. While the industry wants to move quickly to develop 100G DWDM optical network transport, speed isn't the only issue to be addressed. Transmission performance, price, and space and power dissipation per bit also have to be improved over 10G and 40G DWDM transport solutions. In this expert lesson on increasing optical channel rates to 100G, optical expert Eve Griliches, managing partner of ACG Research, looks at the forces driving first-generation adoption of 100G DWDM and examines what the industry is doing to prepare for it, particularly after learning from the mistakes of 40G development. This three-part guide includes predictions on the 100G market.

Table of Contents


Telecom industry prepares for 100G DWDM optical network transport

With the telecom industry giving a resounding "yes" to 100 gigabits per second (100G) dense wavelength division multiplexing (DWDM) optical network transport,

It will take second- and possibly third-generation products hitting the right cost points to move providers from 10G and 40G to 100G.
Eve Griliches
Managing Partner
ACG Research

 the race is on to develop and deliver first-generation 100G DWDM products that will boost optical channel rates to 100G. Having learned from the multimodulation mistakes of 40G, the industry has quickly moved to standardise components so vendors can bring products to market faster and at cost points that will make 100G adoption viable in a shorter time frame.

The first implementations for 100G DWDM network transport will be high-capacity connections between switches, switches and routers, and router-to-router at Internet exchanges and within service provider and carrier points of presence (POPs).Why? One reason is that innovative content providers with huge data centres have put pressure on the industry to develop 100 Gigabit Ethernet on the client side. In addition, the aggregation of 10G links from IP routers has driven the line-side development of 100G DWDM for long-haul optical transmission on service provider and carrier networks.

These innovations should lead to extensive deployment of 100 Gigabit Ethernet in data centre networks, which will be required to manage the bandwidth necessary to aggregate the server and storage community within the data centre. 100G optical transport links will also be deployed as a requirement for higher bandwidth applications and to aggregate internal service elements within the carrier.

100G DWDM and Ethernet standards groups

The following groups are working on relevant 100G DWDM optical network transport (optical channel rates) and Gigabit Ethernet (GbE) standards:

IEEE 802.3 High Speed Study Group (HSSG): Client-side-interface focus for up to 40 km transmission. It supports full-duplex operations and bit error rate (BER) better than or equal to 10 (to the minus 12). With different PHY requirements, a host of physical solutions for 100 Gigabit Ethernet is necessary to support data centre interconnect as well as interoffice carrier connectivity. Support for 10 copper, 100m MMF, 10km SMF and 40km SMF are currently proposed.

ITU Study Group 15; Next Generation Optical and Transport Networks: Focus on metro and backbone DWDM interfaces. Proposals for the next higher bit rate beyond 43G (OTU-4) have been submitted for three times ODU-3 at 130G, as well as a data rate optimised for 100G at approximately 112G.

OIF-CEI 25G Next Generation Backplanes: Defines common electrical interfaces for high-speed applications.

ITU G.709 OTN Standard: Defines basic bit rates, multiplexing and sub-lambda capabilities, as well as how signals are mapped and assigned overhead within transport networks.

OIF DP-QPSK Standard: Defines the modulation scheme for 100G DWDM to speed integrated optical components to market.

Capacity-constrained data centres and long-haul transmission routes will ultimately benefit from the tenfold expansion of Ethernet to 100G. Ethernet data rates have always increased by a factor of 10, while traditional SONET/SDH and optical data rates have increased by four, keeping Ethernet and transport on separate migration paths until they converged at 10 GB/s. This convergence brings up the question of 40G Ethernet and DWDM equality and puts 100G DWDM squarely in the sweet spot for flexibility, cost points and ease of deployment.

Yet 100G transport is not a panacea. There are challenges, such as its susceptibility to transmission problems and the need to meet cost points for wide deployment once it has been tested and deployed. Most likely, it will take second- and possibly third-generation products hitting the right cost points to move providers from 10G and 40G to 100G.

System requirements for 100G DWDM network transport

While high-speed links are needed in wide area network (WAN) and local area networks (LANs), the requirements vary for each topology.

For the LAN, data centre requirements are typically Ethernet-based for 1G, 10G, 40G or 100G Ethernet. In this environment, multivendor interconnectivity is so prevalent that standards are key factors and must be supported in order to interoperate. Often, the interconnectivity is over a single fiber (in most cases, multimode) where spectral efficiency is not important. It is here that the IEEE is working on 802.3 Ethernet LAN standards.

For the WAN and backbone regional networks, spectral efficiency -- a measure of how efficiently a limited frequency spectrum is utilised by the physical layer -- is extremely important. Typically, the International Telecommunications Union (ITU) standard for ITU G.709 is used for the serial rate. The interface is often the proprietary piece. This is where vendors often differ in approach and promote their advantages, which centre in the modulation formats that are chosen, the way dispersion is mitigated on the fiber, polarisation mode dispersion (PMD) mitigation, and the implemented forward error correction (FEC) technology.

So, as higher-speed 100G DWDM network transmission rates are deployed, the following basic requirements must be met by vendors if they want to be considered by service providers:

  • Equipment must be compatible with the existing infrastructure that supports 10G DWDM deployment. Providers are asking vendors to make sure they support the existing regeneration infrastructure in distance and amp length for 10G/40G and 100G. This means equipment must be tolerant to current PMD and nonlinearity and meet the existing engineering rules established for 10G.
  • Equipment must support 50 GHz channel spacing. Since most providers' grids are deployed on 50 GHz spacing, it would be spectrally inefficient to deploy any system on 100 GHz or anything other than 50 GHz.
  • Systems must have a large dispersion tolerance and must be easily deployed and managed. Equipment cannot require tweaking and be difficult when adding lambdas.
  • Products must be cost effective. The target for third-generation 100G equipment is 6.5 x 10G, or 65G of bandwidth at 10G pricing. This goal will most likely not be achieved until the industry deploys third-generation equipment. The estimates are that the price of early 100G deployments will exceed 10 times that of 10G, and second-generation products are expected to top the cost of two times 40G. Clearly, if the price points are there, providers will move faster to consolidate their networks to 100G.
  • 100G DWDM will be deployed first in core and long-haul networks, where distance requirements run from 600 km to 2,000 km and support for fewer than five or six ROADMs (optical filtering) is required. These networks are less cost sensitive than regional or metro networks, where the distance is less challenging and the number of ROADMs increases significantly.

What becomes apparent is that transmission challenges increase with higher bit rates; 40G and 100G DWDM optical networking transport now require advanced modulation and coherent detection techniques to meet network operators' requirements. Most 40G networks use differential phase-shift keying (DPSK) or differential quaternary phase-shift keying (DQPSK) modulation schemes for long-haul networks. For 100G, the expectation is that DP-QPSK will be the standard modulation format used, while individual coherent detection formats will vary from vendor to vendor.

Read the other articles in this Expert Lesson:

About the author: Eve Griliches, managing partner of ACG Research, has extensive experience in technology product management and the telecommunications industry. She was IDC program director for the Telecommunications Equipment group, where she provided in-depth analysis on many key technologies in the telecom market. Her product management experience at network equipment vendors include Marconi (Ericsson), PhotonEx, Nortel Networks, Bay Networks and Welfleet. She can be reached at


This was first published in February 2011