Has optical technology's 13-year victory over Moore's Law come to an end? 
Dana Cooperson
06/06/2008

Moore's Law, originally used to describe semiconductor makers' ability to double chip performance every two years, has been applied to other high-tech markets' ability to exponentially improve performance. In optical communications, for example, dense wave division multiplexing (DWDM) technology has enabled exponentially lower bandwidth costs because it lets network operators send multiple, parallel, gigabit-capacity optical signals down a fibre over great distances with low latency. As network traffic has continued to grow rapidly, the role of DWDM in wireline infrastructure has strengthened as telecoms operators look to carry more traffic at decreased capital cost per unit of bandwidth. As a result, DWDM spend has almost tripled from 2003 to 2007 to US$5.8 billion globally.

For the past decade, Ovum has tracked a key network bandwidth capex metric, cost per gigabit per second per kilometre. As telecoms has moved from an industry measured in 64kbps per voice line to one increasingly denominated in tens or hundreds of megabits per user, optical innovation and vendor competition have spurred transport capex to tumble from pre-DWDM levels of over $2,000 per gigabit per second per kilometre in 1993 to less than $1 per gigabit per second per kilometre in 2007, beating Moore's Law. However, this exponential improvement won't continue without a step-function improvement in system economics that is not expected for another five years.

Multiple innovations, including the optical amplifier and the optical add/drop multiplexer, have been key to the success of DWDM, along with continuous improvements in laser, detector, and filter technology and system software that have allowed ever-higher system capacities and ever-improving operations agility. Relentless competition has allowed continuous reduction in the cost of a 10G wavelength, spurring DWDM to post 13 years of improved cost performance despite periods of relatively low R&D investment in optical systems and components innovation, as the industry collectively licked its wounds from the major market disruptions caused by the explosion of the telecoms bubble.

Reduced R&D spending in the early years of this decade caused one clear casualty: the delayed demand for and supply of cost-effective 40G technology, which is needed to support a step-function in DWDM-based network cost improvements.

The market is finally poised to move beyond 40G early adoption. The best near-term key to achieving another exponential improvement in transport cost efficiencies is 40G-based networks that provide four times the bandwidth with little or no performance compromise over existing 10G networks at less than 4x the capex of 10G-based networks. Ovum is not forecasting 40G to break the magic 4x cost barrier before 2012, but we are hoping to be surprised, given an uptick in investment over the past several years.

So far this year we have seen several promising 40G innovations, available now or expected by year-end, including Nortel's home-grown technology, which offers the best performance on the market and a clear path to 100G; OpVista, which has technology that could have a particularly positive effect on the viability of 40G in metro networks; and progress by Stratalight, Mintera, and others towards standardised 40G modules. Infinera, through technology it will release this year and next, is another vendor on the road to attack cost-capacity-distance stagnation with 40G innovation.

Meanwhile, competitive forces on both network operators and vendors will push innovation in 100G, which in turn will allow the continued success of DWDM in supporting our growing global hunger to communicate economically.

- Dana Cooperson is Vice President, Optical Networking at Ovum.

advertisement