Optical Packet Switching Lives On
Because XLight is making a big thing out of its widely tunable transponders' ability to route packets in their optical form -- something that was considered fairly wild even at the height of the optical networking boom a couple of years ago. Rekindling the idea now -- when the development of less-challenging, all-optical circuit switching has been halted by a lot of vendors (see Lucent Terminates the LambdaRouter) -- seems gutsy to say the least.
To be fair, XLight isn't expecting system vendors to rush out and start making optical packet switches with its transponder any time soon. "The market [for optical packet switching] is not there," says Ilan Brillant, XLight’s marketing director. "On the other hand, we are getting signals that it is not forgotten, particularly among tier ones."
In the short term, XLight is targeting other markets for tunable lasers. “Today our key market is sparing. That's the key market for any tunable laser vendor,” says Brillant. Unlike many, he expects other markets for tunable lasers to pick up momentum (see Tough Times for Tunable Lasers).
XLight’s beliefs appear to be shared by its investors, who are continuing to put money into the startup. Its first round funding of $9 million in October 2000 came from Ascend Technology Ventures, Jerusalem Venture Partners, and Pitango Venture capital. Last month, the same group of investors stumped up fresh funding in what's termed an "internal round." The amount of funding was undisclosed.
There must be something in XLight’s technology for these VCs to continue supporting the company, so it’s worth taking a deeper look at its long term goal of providing the guts of optical packet switches. Put simply, it involves setting up a network of virtual connections where different wavelengths connect different nodes. The tunable laser sits at the periphery of the network, directing packets to their destination by selecting the appropriate wavelength.
XLight's founders developed their ideas about optical packet switching at Ben Gurion University before spinning out the company in 2000. According to Brillant, there are two major challenges -- the transmitter and the detector.
On the transmitter side, the laser has to be able to switch between wavelengths in the course of nanoseconds (ns), so that it can direct consecutive packets of data to different destinations. Brillant boasts that XLight's transmitter can switch between 6,400 combinations of wavelengths in a guaranteed time of 50ns.
Another important requirement, he says, is that the laser should be "dark" while it is switching -- in other words, that it doesn't generate any spurious wavelengths lying between the initial and the target wavelength.
On the receiver side, the detector has to be capable of what's called "burst mode" detection. Packets are unsynchronized, and can arrive at any time, so the receiver's clock needs to be able to "lock on" to the timing of the signal.
Even two years ago, at the zenith of the telecom hype, such a project was considered ambitious (see Researchers Unveil All-Optical Advances). A few startups set out to build optical routers, but they soon crashed and burned. Ilotron was one (see Altamar Buys a Bargain); Luxcore Networks Inc. (formerly Synchordia) was another (see Startup Touts "First Optical Router" and Luxcore Pulls a Switcheroo).
As noted, Xlight is targeting other tunable laser applications to begin with, claiming that its transponder has an edge because it's widely tunable. Note the careful choice of words in XLight's claim. Most vendors make plain vanilla Tunable Lasers, as opposed to tunable transponders, the difference being one of integration. Tunable lasers only contain a light source, which generates a continuous beam of light, and do not include a modulator to encode the digital signal onto the light beam. Transponders are more highly-integrated products that can both send and receive signals, comprising laser, modulator, detector and some electronics to condition and multiplex the electronic signal.
The higher level of integration in the transponder will make it easier for systems vendors to adopt tunable lasers, says Brillant. “To bring tunability into a system today, the system vendor needs a team of engineers to build a transponder around a tunable laser," he says. "With [XLight's] off-the-shelf transponder, the system vendor doesn't have to worry about the optical complexities of tunability."
XLight's transponder comes in a 300-pin MSA (multisource agreement) package, a type that systems vendors will be familiar with.
A couple of other vendors, namely Multiplex Inc., and Optium Corp., have also announced tunable transponders, but their devices are only narrowband tunables.
Multiplex's transponder, for example, covers 8nm of wavelength -- approximately one-fifth of the C-band (1525nm to 1565nm), or 20 channels spaced at 50GHz -- according to Daniel Gay, Multiplex's director of marketing. Optium only offers 8 channels at 50GHz today, but it plans to offer a widely-tunable version before the end of the year (see Kalkhoven Opts for Optium).
— Pauline Rigby, Senior Editor, Light Reading