A fiber-based approach for Ethernet services growth and flexibility/PMC-Sierra offers HyPHY reference design for Ethernet over OTN wireless backhaul

Throughout the telecom industry, there is a fundamental shift from circuit-based to packet-based traffic. Along with this shift in traffic, there is a shift in the transport vehicle from SONET/SDH to Ethernet. The demand for Ethernet links is increasing with Gigabit Ethernet (GbE) and 10GbE, the fundamental units of transport currency.

This article examines the growing demand on service providers for Ethernet transport and how fiber-optic transport platforms can meet that demand and grow with the service provider as its needs evolve. A telco case study is included as an illustration.
A cost-effective Ethernet transport platform

One of the most pressing needs service providers have today is to transport large numbers of Ethernet links and simultaneously accommodate the growing need for 10GbE. Many fiber routes are at exhaust, and some routes with DWDM have few available wavelengths remaining. While service providers are initially looking for cost-effective transport of GbE with a migration path to 10GbE, they are also considering what advantages they could derive from including Ethernet Layer 2 functionality in the transport vehicle. Beyond that, the next challenge will be incorporating leading-edge, fiber-optic capabilities, such as features of the ITU Optical Transport Network (OTN) recommendations in the transport system.

The most cost-effective, fiber-optic transport platform will be the one that can most efficiently use the available fiber or wavelengths for GbE transport. Historically a single GbE was placed on a single fiber or wavelength, but the introduction of muxponders into optical transport systems has significantly increased their efficiency. Muxponders that multiplex eight GbE links onto a single 10GbE wavelength provide an 8:1 improvement in the use of existing fibers or wavelengths.

The muxponder pair in Figure 1 enables the transport of eight GbE links over a CWDM or DWDM wavelength. The traffic from a client on Port 1 is routed to the far-end Port 1. If there were only a single remaining wavelength in the CWDM or DWDM system, the muxponder would allow one GbE to be added immediately and leave room to add seven more GbE links in the future. Muxponders used in a “bookended”  architecture such as this provide quick, easy capacity relief for the growth of GbE links.

Figure 1. Eight GbE to one 10GbE with “bookended” muxponders.

Placing 8xGbE muxponders on each wavelength of an eight-channel CWDM system would provide a total system capacity of 64 GbE links. The total system capacity of a 40-channel DWDM system with 8xGbE muxponders would be 320 GbE links. Future architectures that require massive numbers of GbE links will depend heavily on devices such as the 8xGbE muxponder to get the greatest use possible from the available fiber and wavelengths of CWDM and DWDM systems.
Smooth migration to a Layer 2 network

Muxponder point-to-point systems carrying GbE links can easily be expanded to multi-point networks using Layer 2 functions. GbE muxponders can also be used to build Metro Ethernet Forum (MEF) E-Line and E-LAN services. To make things even easier, the point-to-point muxponder links discussed above can be expanded through software configuration rather than a hardware change-out.

In Figure 2, the point-to-point GbE muxponder link has been expanded to a multi-point network with three nodes. This network is connected by the 10GbE links, each carrying eight GbE links. The high-speed ports serve as trunking ports.

Figure 2. Multi-point network with 8xGbE muxponders.

Layer 2 functionality enables the service provider to make several important enhancements to its network:

VLAN tagging, where any combination of ports on a given muxponder can be associated with any combination of ports on another muxponder;
flexible cross-connect capability for Ethernet virtual channels (EVCs);
E-Line and E-LAN services across the network; and
implementation of rapid spanning tree protocol (RSTP) across the Layer 2 network.

Evolution to OTN functions

The next step in network evolution will be to incorporate OTN functions. OTN is an ITU standard that provides several useful features including:

performance monitoring,
protection switching,
loopback on client ports with diagnostics, and
forward error correction (FEC).

OTN will add new elements to improve performance, reduce cost, and facilitate management in the optical domain without manipulating the client signal—the client signal is transparent. In addition, it will enable in-service performance monitoring and protection switching. OTN features, like the Layer 2 features, can be added to the same hardware platform. The 8xGbE muxponder would then have a 10.709-Gbps OTU-2 line-side interface instead of a 10-Gbps LAN or WAN interface to accommodate the additional OTN performance monitoring and FEC.
A telco case study

A large telco in the U.S. Midwest had strong customer demand for Ethernet services from two mid-rise office buildings that terminate traffic on a metro central office (CO). Fiber links were at exhaust between the CO and the office buildings. Both office buildings were communications hubs with significant traffic originating from and terminating in them. Several of the telco’s large corporate customers had offices in both locations. In addition, several of these customers wanted Ethernet VLAN service between both offices and the terminating CO.

It was important to provide the GbE links between the office buildings and the CO, but it was also important to deploy a platform that could provide a migration path to Layer 2 services, such as VLANs, and possibly OTN in the future. In addition, the telco wanted a 10GbE path between the two office buildings for diversity and also to facilitate VLAN service with RSTP.

The telco decided to place DWDM on these links and overlay the DWDM with 8xGbE muxponders to conserve wavelengths in anticipation of the growing demand for GbE at each office building. The 8xGbE muxponder at each location provided an immediate solution for point-to-point Ethernet connectivity, and the diverse path between the two office buildings provides protection in the event of a fiber cut. The three-node network shown in Figure 3 can easily be upgraded with a software configuration change to facilitate VLAN services as they are requested by the telco’s customers.

Figure 3. Telco network with DWDM and 8xGbE muxponders.

Another feature of the 8xGbE muxponder that was attractive to the telco was that the GbE ports are SFP based. As GbE links are required, SFPs are added to each muxponder, as shown in Figure 4.

Figure 4. SFPs being added to a muxponder.

Low-cost 1310-nm, short-reach (or 850-nm) SFPs can be used in the muxponders to interface to the GbE equipment located in the CO or office buildings. The initial cost of an 8xGbE muxponder will be higher than that of a single-channel GbE transponder—but the cost of a fully loaded muxponder will be significantly lower than that of eight GbE DWDM transponders (Figure 5). The cost of seven additional fiber pairs or seven wavelength pairs on a CWDM/DWDM system for single-channel transponders would be an extra expense.

Figure 5. GbE links using individual DWDM transponders and wavelengths.

The 1310-nm, short-reach GbE SFPs typically have short lead times and can be stocked for use on all GbE muxponders, wherever they are deployed. They cost significantly less to carry in inventory than single-channel transponders.

The telco concluded that a muxponder was a strategic advantage because demand can materialize seemingly out of nowhere and the time lost in ordering and placing equipment or—worst case—placing fiber or leasing dark fiber equates to lost revenue.
Conclusions

The telco in the case study found that the building of a multi-node network using 8xGbE muxponders deployed over DWDM had many advantages. It provided a cost-effective way of delivering a GbE facility to its growing base of business customers. GbE links can be added very quickly by adding a GbE SFP to an existing muxponder at each point on the link, allowing faster time to market and revenue. The diverse routed network can be enhanced via software upgrades to allow Ethernet Layer 2 functionality and, in particular, E-LAN-type service. Down the road, the telco may consider an upgrade to include OTN features that will provide improved reach through FEC and additional operations, administration, and maintenance (OA&M) functions. It has proven to be a very flexible platform with an easy migration path.

APRIL 19, 2010 — PMC-Sierra Inc. (Nasdaq:PMCS) has introduced what it asserts is a complete reference design for its HyPHY chipset that supports standards-based transport of Ethernet-based services via Optical Transport Network (OTN).

ITU-T G.709 version 3 (10/2009) specifies direct mapping of Gigabit Ethernet (GbE) to the OTN layer by defining a new container for the smallest Optical Data Unit (ODU), ODU0 (1.238 Gbps), ensuring complete timing transparency and bandwidth efficiency of Ethernet-based client signals. In conjunction with HyPHY’s support for timing synchronization over packet services, the reference design enables cost-optimized Layer 1 OTN networks with full support for Ethernet network timing distribution, PMC-Sierra says.

PMC-Sierra’s HyPHY chipset integrates support for mapping and multiplexing of GbEs into sub-ODU1 (2.488-Gbps) containers. The new PM5423-KIT ODU0 reference design leverages HyPHY’s flexible packet interfaces and OTN Payload Tributary Mapping (OPTM) technology to add support for the newly defined OTN mapping modes, including:

transparent GbE mapping and de-mapping into/from ODU0 via GMP
ITU-T ODU0 multiplexing and de-multiplexing into/from higher order ODUks.

PMC-Sierra’s ODU0 reference design can be used with either the PM5420 HyPHY 20G or PM5426 HyPHY 10G device. It includes what PMC-Sierra describes as a complete collateral suite, including a hardware reference design for evaluation, design specification documentation, and FPGA source RTL for integration.

The HyPHY 20G is designed to deliver high-capacity framing, mapping, and multiplexing of Carrier Ethernet, SAN, OTN, transparent bit services such as video, and SONET/SDH to enable carriers to reduce the number of network elements, while incrementally adding bandwidth or changing service mix per node without forklift upgrades. The platform provides client-agnostic, rate-agile SFP and 10-Gbps pluggable (XFP or SFP+) interfaces and a set of flexible system interfaces for networking to a variety of switch fabric architectures.

The HyPHY 10G provides a lower density option for optical platforms.

The PM5423-KIT ODU0 FPGA reference design is available today. The PM5420 HyPHY 20G and PM5426 HyPHY 10G devices are also available.