Several factors shape the new optics landscape
by Sinclair Vass
It is no surprise to the optical communications industry that bandwidth demand continues to soar and that we have only just begun to tackle the challenge of how to expand and optimise the optical network to support that growth.
There are an increased number of users and ways for them to access the Internet. On the consumer side, applications like My Space, YouTube, and iTunes are generating an increasingly higher use of bandwidth and triggering unpredictable traffic surges in the network. On the enterprise side, computing system throughput is doubling about every two years, according to the IEEE Higher Speed Study Group. This is driving the need for high-speed connectivity between routers and out into the wider network.
End users want and expect instant access to online applications, and that expectation will only continue to grow and become the norm as younger generations mature and think up new ways to leverage the online and use of online applications is often limited by the network's and on-demand world.
Today, more pervasive use of online applications is often limited by the network's ability to support those applications. The challenge for network equipment manufacturers (NEMs) and service providers is how to design next-generation networks to effectively manage unpredictable, IP dominated, high-bandwidth traffic while simultaneously managing relentless pressure to reduce network equipment and running costs as bandwidth becomes ever cheaper to the end user.
So what is in store for the future? What key trends in optical communications are going to contribute to the evolving on-demand world?
Network traffic has long since become "data dominated," and service providers are evolving their networks to utilise the power of IP, usually running over an Ethernet transport layer, to efficiently handle data packet transport and routing in long-haul and metropolitan-area networks. However, key to the effective management of optical bandwidth is an agile and optimised WDM transport infrastructure that permits network operators to remotely provision their network according to local bandwidth needs.
With a fully agile WDM optical layer, the service provider can simply use his network management software to remotely add, drop, or re-route an optical wavelength between any two points in his network to relieve a local bandwidth bottleneck, without compromising any other part of the network in the process. In this way, the service provider can provide a consistently high quality of service to his customer base, and differentiate customer billing tariffs based on the guaranteed bandwidth provided.
Network management costs are also drastically reduced. A wavelength upgrade used to take several days and involved multiple truck rolls to local points of presence to reconfigure terminal equipment. Such upgrades now only require just a few clicks of a mouse in the central office.
The business case for WDM is compelling. As this becomes the configuration of choice to support network transport and routing—a recent Dell'Oro report stated that sales of WDM equipment in the first quarter of 2008 overtook SONET/SDH for the first time ever—this in turn drives the need for a new and highly intelligent breed of agile optical components that operate at the very heart of these long-haul and metro networks.
In addition to support of remote reconfigurability, these components must be able to automatically detect and adjust to changing optical link loadings to keep the network balanced at all times. They require optics vendors to invest a great deal of technology expertise and R&D resources during the development process, as well as efficiently produce high volumes of the products to support rapid network deployments at an attractive price point.
Some of the key optical elements critical to an optimised and agile WDM transport layer include the following:
Tunable XFP modules. Tunable lasers provide dynamic reconfigurability by allowing network operators to remotely switch from one wavelength to another on-demand. They also ease the cost of purchasing, storing, and managing spare devices for wavelength management. The tunable XFP module is becoming the tunable device of choice because of its very small footprint, high performance, and ability to support faster network speeds.
Optical vendors must make a large investment to integrate all of the optical functions associated with an optical 10-Gbits/sec laser, modulator, and amplifier onto a single multi-stage indium phosphide substrate. Strong partnerships with ASIC vendors are also needed to shrink all of the electronic control functionality into the size and power consumption constraints of the XFP form factor.
Wavelength-selective switch (WSS) components. A new generation of optical switches that can route one or several chosen wavelengths from an input to any output port enables adding, dropping, or re-routing channels at an optical node to support multiple ring or mesh-based architectures. The technology currently uses free space MEMS arrays (essentially rows of electronically steerable mirrors) that require significant technology know-how to fabricate and stringent quality testing to ensure reliable operation at mission-critical points within agile optical networks.
WSS components are prevalent today in long-haul networks in the U.S., but are increasingly being used in EMEA to support new long-haul and metro network deployments. Simpler flavours of WSS with lower channel counts are also being developed to support the demanding price points needed for deployment in the metro/access environment, pushing flexibility right to the edge of the network.
Monitoring, multiplexing, and amplifying subsystems. Components that enable a flexible and agile optical network to remain in balance are also crucial. Since the number of wavelength channels in any optical link is essentially unpredictable, networks require devices that can intelligently and quickly react to traffic overloads and underloads to prevent network instability. Optical wavelength monitors, dynamic wavelength-flattened amplifiers, and active dispersion compensators are all examples of this new generation of optical components.
A holistic approach called "functional integration" factors size, cost, power efficiency, and increased performance into the development of new products to meet such requirements. This could involve more optical functions on the same substrate (such as photonic integrated circuits), multiple components inside a module, or even multiple optical functions on single card. An example of this high-end integration are "super transport blades" that have all the main amplification, monitoring, and switching functions of a typical ROADM combined on a single card, with a backplane interface to the NEM optical equipment. Such functionally integrated subsystems allow the NEMs to provide more value and systems that are easily integrated into service providers' existing infrastructures.
The content explosion happening over the Internet is also pushing the need for higher data rates of 10-, 40-, and eventually 100-Gbit/sec speeds through network pipelines. Today, 40-Gbit/sec capabilities are almost always a requirement when NEMs are bidding for service provider contracts, and many new networks can already be seamlessly upgraded with 40-Gbit/sec links when bandwidth exhaust starts to occur. U.S. network deployments at 40 Gbits/sec are leading the way, with Europe gaining traction, particularly in densely populated regions.
The drive to 40 Gbits/sec is not only occurring in the long-haul transport network. Increasingly, 40-Gbit/sec router equipment is being deployed in the LAN to support the data traffic generated by businesses and local campuses, which in turn is driving the need for short-reach 40-Gbit/sec optical connectivity. It is only a matter of time before the access and regional metro networks start to feel the strain from this increase in LAN/WAN traffic and also require an upgrade to higher line rates.
In terms of 40-Gbit/sec network equipment, first-generation platforms have generally been expensive with challenging cost structures. They were either internally developed by NEMs from immature optical components or purchased at a circuit-pack level from one of the few 40-Gbit/sec start-up companies. These systems are not easily scalable to support high-volume demands.
Next-gen 40-Gbit/sec implementations are based on more standardised approaches. These new systems are designed to meet the cost and volume requirements of service providers and facilitate widespread deployment in the network. At the module level, the key component is a 300-pin transponder that converts traffic signals from electrical to optical formats (and vice versa) as signals are added and dropped in 40-Gbit/sec networks.
It's important to emphasise that 40-Gbit/sec transmission issues cut across all points of the optical network. Customer choices on 40-Gbit/sec transmission technology affect their choices for WSS technology, for example. Advancements in modulation formats and optical communications components must continue to support faster network speeds, making the technical know-how of vendors and joint development even more critical.
And while 100 Gbits/sec is definitely the hot industry topic, there aren't yet any mature products for such speeds other than some parallel optics components using ribbon fibre to transport 10Ã�10-Gbit/sec data streams. The most likely first commercial application of 100 Gbits/sec will be for point-to-point LAN interfaces up to 300 m to support large router traffic, with first deployments occurring in late 2009. Client-side applications of 100 Gbits/sec for shorter inter-office connectivity will occur later in 2010, and the IEEE is working on both standards right now.
Line-side applications for longer reaches over the Internet will take more time. This is the most technically challenging implementation and will likely employ novel polarization multiplexing techniques to break the 100-Gbit/sec signal into several lower bit rate and co-propagating data streams.
As 100 Gbits/sec becomes a reality, 40-Gbit/sec structures will not disappear but will continue to be a cost-effective option for service providers for some time to come. Effective transport of native 40-Gbit/sec traffic over a 100-Gbit/sec link is an active topic for the standards bodies, including the ITU.
In an environment where unpredictable, high-bandwidth applications can rapidly drive network overload, NEMs must support service providers with fast deployment of new customised equipment and upgrades. As a result, optical component and module vendors must support the NEMs with aggressive lead times and scale throughput quickly in response to upside demand.
An aggressive lean manufacturing strategy drives waste and cost out of the supply chain. Using demand-pull techniques originating from the high-volume automotive industry, along with the aggressive selection and management of the supply chain at all levels, optical component vendors are driving lead time and product realisation costs down.
Three-week lead times have become the norm for major optical vendors, quite an achievement given the very limited ability to forecast the volatile demand for optical components worldwide. Only the bigger vendors have the resources and power to do this effectively, and supplier consolidation will happen over time as the NEMs demand higher service levels from their preferred vendors.
For the next generation of optical components and modules, product development teams are being measured on their ability to launch new products that support lean manufacturing principles on day one, and that also meet the demanding cost requirements of an industry where significant year-on-year price erosion is the norm.
Strong subcontract manufacturing partnerships also remains important as "build in-house or outsource" decisions depend on many factors, including the need to protect internal intellectual property, the volatility of the demand profile, the labour content of the manufacturing cycle, and the proximity to the end user.
The sense of urgency in the optical communications industry must focus on the business model. Without a consolidated, robust, and financially healthy supply chain, industry innovation and advancements cannot continue to progress. New and exciting Internet applications will stall or will be delayed as the optical transport infrastructure struggles to keep pace.
Optical component vendors also need to have an unrelenting focus at each level of the supply chain, as driving profitable business models is crucial. Only the vendors that can quickly streamline internal processes to support high-volume lean manufacturing and assembly, fast turnaround times for product delivery, and reduced lead times will survive and thrive in the optical communications industry.
Sinclair Vass is the EMEA sales director, communications, at JDSU (www.jdsu.com).