Telecom goes Terabit at OFC/NFOEC 2008
SAN DIEGO, CA—I still remember the excitement back in 1992 when, as a product/marketing engineer at Corning Incorporated (Corning, NY), our group had one million dollars in backlog of fused biconic taper couplers with high hopes that fiber-to-the-home was just around the corner.
SAN DIEGO, CA—I still remember the excitement back in 1992 when, as a product/marketing engineer at Corning Incorporated (Corning, NY), our group had one million dollars in backlog of fused biconic taper couplers with high hopes that fiber-to-the-home was just around the corner. Who knew that it would take the telecom downturn of late 1993 and the agonizing ‘bubble’ of 2001/2002 before fiber-to-the-home (FTTH) started to look promising? But now that FTTH and 40 Gbit/s systems are being commercially deployed (see www.laserfocusworld.com/articles/318567 & 318568 & 309265), it seems that the telecom industry is already looking ahead to Terabit networks to feed what is finally real and tangible bandwidth demand.
Feeding the bandwidth beast
Evidence of this growing need for bandwidth proliferated in both Plenary and Market Watch presentations at this year’s Optical Fiber Conference/National Fiber Optic Engineers Conference (OFC/NFOEC), which ran from February 24–28 in sunny San Diego, CA. In his Tuesday Plenary session on “Managing Proliferating Traffic Growth”, CTO of Qwest Communications Pieter Poll said that Qwest’s Internet traffic is doubling every 16 months, from levels of 9000 Terabytes per day in 2007. As the industry anticipates moving from 10 Gbit/s to 40 Gbit/s networks, Poll said that his company is now considering a direct move from 10 Gbit/s to 100 Gbit/s networks and is hoping to see a 100 Gbit Ethernet (100 GE) standard by 2009.
Still more phenomenal bandwidth demand was evidenced in Wednesday’s Panel III Market Watch entitled “Enterprise: Will the Next Speed Jump Bring a Boost for Optical.” Network Architect Donn Lee from Facebook presented a strong case as to “Why 100 G Pipes Aren’t Enough for Core-Rich Datacenters” when he explained that Facebook has 66 million active Internet users and is adding 1 million users per week. Calling Facebook the #1 photo site on the Internet and the 6th busiest, Lee said that his company needed 100 Gbit/s core switching last year to keep up with growing demand, and is filling the gap today by adding multiples of 10 GE networks connected by fiber in the core switching systems for Facebook. Lee’s presentation was followed by Robert Schrage, head of network architecture and standards at Reuters, who also described the proliferation of Internet demand for its news and marketing services, noting that Reuters financial data is updated over 8000 times per second. His presentation, with the same title as the Panel III overall title, described the coming Terabit network and how new optical technologies will need to be developed to overcome the N x 100 GE bottleneck that is already being anticipated by pushing the limits of legacy installed optical fiber.
This year’s OFC simply leapfrogged last year’s 100 GE discussions and moved directly to Terabit networks. The second Plenary presentation “Toward Terabit Ethernet” from Bob Metcalfe, general partner of Polaris Ventures (who is credited with helping to invent Ethernet), described the “it” and the “they” within the phrase “build it and they will come.” Metcalfe describes “it” as Terabit Ethernet, and “they” as the bandwidth-hungry voice, video, and data users with ever-increasing traffic demands. Among his many other revelations, Metcalfe calls dense-wavelength division multiplexing (DWDM) a “silver bullet” that helped communications providers respond to rapidly growing bandwidth demand by simply multiplexing the capabilities of the existing fiber infrastructure; however, he cautions that the Internet isn’t done yet, and something more will be needed to bridge the next bandwidth gap.
Metcalfe’s presentation largely discussed the demise of synchronous optical network (SONET) architectures, and joked that in the “secret” Ethernet meetings that he attends that conspire to kill SONET, he sees Ethernet winning for a host of reasons that include its reliance on standards, its 10X per generation speed growth compared to SONET’s 4X per generation speed growth, and most importantly, its multi-vendor interoperability and tolerance to “alien” wavelengths—a reference to the ever-shrinking channel spacings in DWDM networks. Beyond DWDM, Metcalfe predicts that Terabit Ethernet will be chaotic, and may require thinking beyond 50 GHz grid spacings and 1550 nm wavelengths. Metcalfe ended his presentation with a plea to the audience: How will the Internet solve the energy crisis? He hinted at a world that fuses photovoltaics with optical communications—after all; both convert photons to electrons. While better communications may mean less travel, and more efficient travel to save energy, Metcalfe cautions that “we’re not going to get to Terabit Ethernet with the current infrastructure,” a strong hint that we all need to look to new materials, new sources, and new means of moving photons around.
Although attendance was about the same as the last few years (approximately 12,000), OFC/NFOEC 2008 saw the highest level of technical paper contributions since the peak in 2003. Paper submissions increased 20% this year to 1155 total—with 40% coming from the Pacific Rim, and more than 20% coming from Europe. This strong international presence was also reflected by the fact that of the 600 exhibitors, roughly 40% were from outside the U.S.
As usual, the 35 accepted post-deadline papers (from 114 submissions), did not disappoint. Highlights included a record spectral efficiency of 4.2 bits/second/Hz from a group at AT&T Laboratories, and a record capacity-distance product of 41.8 petabits/second-kilometers from the French side of Alcatel Lucent. Corning reported on a new fiber that can be bent to a radius of 5 mm—an important improvement in FTTH systems where smaller-bend-radius fibers mean smaller boxes on the sides of houses, and higher tolerance for installers who tend to bend fibers too much. And of course, progress continues in the march towards monolithic integration of some impressively complex systems, including a 10-channel by 40 Gbit/s differential quadrature phase-shift keying (DQPSK) transmitter from Infinera. And echoing the mention of a revival in coherent receivers for optical networking in the Plenary presentation from Herwig Kogelnik, adjunct photonics systems research VP of Bell Labs, Alcatel-Lucent, a few post-deadline papers hinted at the resurgence of coherent transmission for high-capacity systems. Dumped two decades ago when erbium amplifiers arrived, coherent transmission may be viable for practical use now that narrow-band lasers are available and techniques exist for extracting the carrier signal at the receiver.
Did you ever wonder what happened to tunable swept-laser technology from Iolon? Look no further than Luna Innovations (Blacksburg, VA), who purchased the assets of Iolon from Coherent and is now marketing its Phoenix 1000 MEMS-based, external cavity laser based on the former Iolon ‘Apollo’ class of tunable laser. This compact device can be used for applications in communications, fiber sensing, spectroscopy, and metrology, by providing fast tuning up to 500 Hz over the C-band wavelength from 1515–1565 nm with linewidth less than 2 MHz and excellent phase-noise characteristics.
Finisar (Sunnyvale, CA) unveiled three new technologies at OFC/NFOEC: A 17 Gbit/s 850 nm VCSEL for the emerging 16G Fibre Channel standard; a 200 km 10 Gbit/s DWDM XFP transceiver module based on Finisar’s chirp-managed directly modulated laser (CML) technology; and a 40 km reach transceiver module. In addition, Finisar announced that it is partnering with Nistica (Bridgewater, NJ) to offer reconfigurable optical add/drop multiplexers (ROADMs). Finisar also partnered with Vitesse Semiconductor to demonstrate a 300 km optical link at 10 Gbit/s using standard non-return to zero (NRZ) modulation that does not require dispersion compensating fiber (DCF).
The single-photon source operating on the show floor from Quantum Communications Victoria (Parkville, Australia) was noteworthy, considering that these usually have to be operated in dark rooms to avoid noise. A patented technique for optical pumping of single-atom nitrogen defects in diamond creates an excited state that generates true single photon emission at 700 nm, allowing show-floor operation, said QCV development engineer Brant Gibson.
For DWDM spectral analysis, Lightwaves2020 (Milpitas, CA) announced a new Optical Channel Monitor (OCM) based on its proprietary optical thin film technology and innovative micro optics. The OCM can scan the wavelength range of C-, L- or C+L bands sequentially and measure or monitor the optical channel power and wavelength at ITU grid spacings of 50 or 100 GHz, and has the unique capability of measuring/monitoring multiple wavelengths simultaneously.
And finally, newcomer Lightwire (Allentown, PA) introduced the first CMOS Multi-Source Agreement (MSA) standards-compliant 10 Gbit/s Ethernet (GE) SFP+ LRM (small-form-factor pluggable long-reach multimode) transceiver that integrates light-modulation circuitry into silicon—perhaps one of those new ‘all-optical’ technologies that begin the drive towards Terabit networks.
—Gail Overton and Jeff Hecht, Boston Correspondent: New Scientist magazine and Contributing Editor: Laser Focus World magazine