The Itch to Switch
123jump.com
Feb 06, 2001, 11:57 AM ET
By Alexander Vantchev
Counting Sheep Before The Photonic Dream, or How Do You Flick a $3.25–Billion Light Switch
Speed is marketable. The Internet is, among many other things, slow. Faxes are messy, e–mails have to be typed and replied to; the pet peeves of the information age haven’t changed a whole lot since the fax machine first squeaked a digital carrier tone on a regular phone line a couple of decades ago.
'Broadband boom' expectations were reasonably triggered by the common logic that, at current levels of development and growth, technology is bound to meet need sooner rather than later. Efforts to break the speed limits were directed at all networking segments – transmission, switching and routing, network management.
The evolution of fiber optics, and advent of laser transmitters and receivers made ultra–fast transmission possible. But, one of the biggest challenges then became to switch and direct the data streams traveling by light, without converting them to electronic signals and then back to light – an unpleasant 20 times speed bleed.
The dream of the all–optical, or photonic, switch was born.
CIR Inc. expects the U.S. market for optical switching and subsystems to grow from $247 million in 2000 to $2 billion by 2004. According to Pioneer Consulting, the optical switching market in North America will grow from $427 million in 2000 to over $10 billion in 2004. The forecast for Europe is growth from $116 million to over $4 billion within the same time frame.
Two years ago, on February 23, 1999, another pioneer, the hallowed Bell Labs, brain of giant Lucent (LU), announced its 'seesaw' switch – an all–optical solution, based on micro–electro–mechanical systems (MEMS) technology – which under a microscope looked like some chunky flap on hinges and rails. A golden mirror was used to swing and direct the beam down the proper fiber.
Since then, MEMS were thought to be the answer to all–optical switching. Bell Labs was fast to develop its MicroStar technology, allowing Lucent to market the WaveStar LambdaRouter 15 months later. The device was touted as the world's first high–capacity, all–optical router with more than 10 Terabits per second total switching capacity, commercially available in December 2000.
The first big carrier to deploy the tera–gadget was Global Crossing (GBLX) in its global optical network spreading to more than 200 cities on five continents.
It was a groundswell of firsts. Both startups and established players were announcing their first all–optical solutions in all sorts of ways. Companies like Tellium even used the word twice: first–to–market and first–in–service for its Aurora Optical Switch at 1.28 Terabits per second.
Meanwhile, Nortel (NT), Lucent's biggest competitor in the optical transmission field, coughed up $3.25 billion for the development of its all–optical MEMS switch. Note that this money – which Nortel paid for the acquisition of just one switch–making company, Xros – is 1.6 times bigger than the whole estimated U.S. optical switch market in 2004, and over 13 times bigger than the market in 2000. The first beta tests of the 3–billion baby were done in December 2000.
One thing is certain about MEMS technology – it is simple and popular. MEMS are actually new to optical switching. They have been used in car airbags, ink jet printers, blood pressure sensors, and recently, in camcorders, DVDs, and many other consumer products. According to a recent survey by Cahners In–Stat Group, the consumer market for MEMS will grow from $200 million in 2000 to over $1.5 billion in 2005.
It is clear that money has poured into MEMS, but is there any money left in it?
MEMS – a Mammoth Already?
Nope, MEMS is totally passé – according to Timothy Cahall, CEO of Columbia, MD–based Trellis Photonics, which has its development and engineering centers stationed in Israel. A true lab–to–fab company, sprung from the Hebrew University of Jerusalem, Trellis is aggressively promoting its own first: the Intelligent Lambda Router, "the industry's first truly all–optical switching fabric."
Cahall thinks that the $5 billion already invested in MEMS for telecommunications are "speeding their way to money heaven," a total waste. What Bell Labs announced as a "disruptive technology" that "changes the paradigm for an entire industry," and "will form the basis for switching technology that will be used in future networking systems," to Cahall is "a true tragedy," which "lies in the delay and disillusionment that this waste will cause in achieving the all–optical network."
Breaking the Waves
Cahall finds the big flaw of MEMS to be their inability to manage the power level of individual wavelengths, which means that they need a very careful (read: expensive) balance, or one "misbehaving" wavelength at too high a power can interfere and effectively shut down all of the wavelengths in every fiber that the signal passes through, and thus cause a blackout in large portions of the Internet.
Cahall thinks that these problems throw the industry in vicious circles of patches over patches and are forcing it to dump "top notch" equipment every two or three years.
Photonically Phantastic
Trellis, of course, has a switch pitch of its own. Its technology is based on a novel photorefractive material – potassium lithium tantalate niobate – with optically induced diffraction gratings, an all–solid–state device using electro–holography.
No smoke and fluttering mirrors, photonically phantastic, and, according to Trellis, ready for the world in mid–2001. The company’s future facilities are funded with $25 million of venture capital and are expected to open 650 jobs in the optical haven of Howard County, MD.
The Quiet Side of the Spectra
In 1996, in a private home in Santa Rosa, CA, the company SpectraSwitch was born. In contrast with the radical novelty of Trellis, it built its all–optical switches with something unbelievably old and popular – liquid crystals. The familiar stuff in all those digital watches, displays, etc.
China Development Industrial Bank (CDIB) and Nippon Investment & Finance Company Ltd. (NIF) provided the initial funding.
SpectraSwitch places its bet on the mature fabrication infrastructure of liquid crystals to allow fast and reliable mass production of its WaveWalker family, which it hopes will walk steady on the scene where the runners are starting to wheeze.
Apart from their announced readiness for the high bit rate of future photonic networks, WaveWalkers are really neat looking widgets and have been attracting quite a lot of attention in the industry recently.
Where Startups Dare
SpectraSwitch is an interesting startup, but liquid crystal switching technology is not a new thing, it just used to be slow and unstable. Recent refinements, though, seem to have attracted the attention of such optical heavyweights as Corning (GLW), which use it for its PurePath family of optical switches.
One other household item, namely the ink jet print cartridge, gave birth to another optical switching idea – the "bubble" switch of Agilent (A) – unsurprisingly, a spin–off of ink jet king Hewlett–Packard (HWP). When a light beam must change direction, a microscopic bubble pops up on its path and 'breaks' it.
Show me the Money
Whether it's light, electricity, or pigeons, the information delivery business is to get the message from point A to point B as fast as possible, with as little wrinkles as possible, and to make yourself and your investors some profit. If it moves the terabits right, it may be pigeons for all carriers care. And some companies prefer to have the word 'Terabit' in the name of their product, and not 'light', 'photonic', or 'all–optical'.
It is of little importance how often the light traverses electrically, if customers are big, many and happy. Market share is won on the market, not in the lab, and all photonic reveries may never leave the field of dreams, if they don't prove their worth in the real world.
Companies like Ciena (CIEN), Cisco (CSCO), Sycamore (SCMR) and Tellium relied on Optical–Electronic–Optical (OEO) solutions and got their terabits – and orders – moving. The mature electronic processing, powered by parallel computing, which doubles or triples speeds and reduces costs each year, is still a viable option on today's market. Most people access the net with modems and LAN–cards, after all.
As interesting as the scientific showdown is, the competition of business models may ultimately decide the score.
One of the promising players, Avici Systems (AVCI) recently deployed its Terabit Switch Router (TSR) in the West Coast backbone of "SuperNet" – an ultra–high bandwidth project, run by the National Transparent Optical Network (NTON) Consortium, which includes Nortel, GST/Time Warner Telecom (TWTC) and Sprint (FON) among others, under the auspices of Defense Advanced Research Agency (DARPA). Also known as Next Generation Internet (NGI), its purpose is to develop the Internet applications of the future, where terabits of data are exchanged real–time.
Apart from running on the leading edge, Avici has a growing customer base in North America, Europe and Asia. Its director and chief operating officer Steve Kaufman was president of Lucent's multi–service core networks division, and general manager of its microelectronics division. Nortel owns 14% of Avici.
Avici's jewel – the Terabit Switch Router – was demonstrated last April, in SUPERCOMM 2000, Atlanta, GA, paired with the all–optical suite of another high–profile optical company – Corvis (CORV).
'Fess Up, Corvis!
Some time last year, Corvis was probably the most likely answer to the question: "who has the hypest all–opticals of them all?"
Dr. David Huber, founder of Ciena, is founder and CEO of Corvis. He controls about 28% of its stock. Cisco owns some 5%. Kleiner Perkins Caufield & Byers (KPCB) own about 10%. KPCB has backed many optical startups, including Cerent Corporation, which Cisco bought for a record $7.4 billion in 1999.
Recently KPCB founded iolon, a company with a similar optical profile, where Optical Capital Group (OCG) is also an investor. David Huber is an investor in OCG, according to Light Reading.
A Columbia, MD–based company, symbolically residing on Albert Einstein Drive, Corvis made a breakthrough in optical transmission when Williams Communications (WCG) effectively transmitted multiple 2.5 Gbps optical signals across more than 3,200 kilometers without electrical regeneration using Corvis' integrated solutions on July 11, 2000. 16 January 2001, Williams extended its two–year contract with Corvis from $200 million to a multi–year $300–million agreement.
A similar feat was accomplished by another $200–million client – Broadwing Communications (BRW) transmitted 10 Gbps signals from Phoenix to Forth Worth and back to Phoenix (a good 4,000–kilometer section) without electrical regeneration using Corvis' OC192 product.
Four thousand kilometers is far, even for a trial test, which it was. So far, so good, but investors are getting jittery. They remember Corvis' crazy IPO which cha–chinged $28 billion into the company's war chest on day one.
Since then a lot of things have changed. Radical voices challenge the MEMS' dominance, mainly from the private sector, but it was public companies' stock, which lost a good deal of investors' money in 2000, and losing money is not a good deal. Other startups claim they have equal or even better all–optical systems, England–based Ilotron, to name one.
Analysts ask Corvis constantly to disclose its technology, because after a 75% stock slump, the name and fame of Dr. David Huber alone may not be enough to sustain confidence. Good news: on January 25, 2001 Corvis announced it is the first company to recognize revenue on a commercial deployment of an all–optical core switch. Bad news: still only three customers: Williams, Broadwing and Qwest (Q), and the Qwest $150–million deal is a bit iffy.
Still, Corvis maintains the viability of its business model. It sells an end–to–end all–optical system: transmission, switching and network management all in one package, or rather black box – if it goes down, you'll see what's in it.
Some see the legendary Corvis secrecy as a way to maintain clients' dependability in a highly competitive market, but the feeling of dependence is also not generally pleasant.
Another explanation offered is the 180–day lockup period for insider shares, which expired on January 23, 2001. Maybe with the shares unlocking, Huber will share some info. on his all–optical technology.
After all, it's all about shedding light, isn't it?
The Smart Part
There is still something fishy about that all–optical thing. In every technology mentioned, there is an electric current somewhere, and it is actually driving the light. MEMS are electro–mechanical; in a liquid crystal switch, the beam is polarized to change direction in the beam displacer, and this polarizing is done by applying voltage to the crystal cell; the bubble in the bubble switch is popped–up electrically; the Trellis router uses electro–holograms, etc.
There is no 'unplugged' light technology, yet. There is always some chip somewhere – which does the 'smart' part – that is reading information about where the light wave is supposed to go, and then another electrically driven part is doing the switching. Without electricity, the light switch is useless ... as many California residents have come to realize.
There are those who say that calling all this "all–optical" is wishful thinking. In a true all–optical system, light will control light, photons will open and close the gates, not electrons, and this idea is far from new, too. In fact, Bell Labs announced the development of photonic switching in 1991.
But to process a light signal with light will require photonic 'transistors', and photonic 'chips'. In 1990, Bell Labs announced its 'optical digital processor' – an invention clearly designed to address these issues.
Since then, not a lot has been heard about the optical computer, but in July 1999 privately held Cyber Dyne Computer Corporation announced that it hired a director of new product development, who was a former Bell Labs employee. Cyber Dyne's products include photonic transistors, light speed routers and switches, mass storage devices for over 500 terabytes of data and other sci–fi sounding devices.
The same San Diego–based company later reappeared as AON Corporation – AON stands for all–optical–network – with the same product gamut. Its integrated circuit (IC) solution is holographic – the same thing Trellis uses for its Intelligent Lambda Router.
If we believe a recent announcement by the Semiconductor Industry Association (SIA) that the worldwide chip market eclipsed $204 billion in 2000, the question of where the future photonic chipmakers are heading is a no–brainer.
The need to switch light may bring the itch to switch to light.
*******************
Jack |
