Old Switch, New Tricks
Previously, Lynx had designed its subsystems around the C- and L-bands (1520 to 1565 nanometers and 1570 to 1610 nm, respectively) used for long-haul Wavelength Division Multiplexing (WDM) communications. The improved switch design addresses a much wider range of wavelengths in one device -- from 1265 to 1610 nm -- taking in applications at 1310 nm, which is often used in enterprise networks, as well as CWDM, a cheaper version of DWDM based on widely-spaced channels.
"Every market uses different wavelengths," says Steven Montgomery, president of ElectroniCast Corp., a market research firm specializing in optical components. "The wider you can go, the more customers you can address."
Seeing that the markets for long-haul gear in general and optical switches in particular seems to have stalled, it doesn't take a genius to work out that Lynx needs to find new opportunities in order to ensure its survival. Just a few weeks ago, the company laid off 23 percent of its staff (see Headcount: Scary Times Continue ).
To its credit, Lynx appears to have solved one of the key limitations of PLC-based components -- namely, wavelength sensitivity. This is caused by the fact that switching in PLCs (planar lightwave circuits) is done via a change in the material's refractive index, which by nature is wavelength dependent. Lynx isn't saying exactly how it achieved this breakthrough, however.
One of the first areas in which Lynx expects to apply the new technology is protection switching. Lynx introduced and demonstrated a prototype of its Photon.PROTECT 1:N protection switch at the ECOC show just a few months ago (see Lynx Intros Protection Switch).
The protection function was originally designed for a client in mainstream telecom, says Michael Leigh, president of Lynx. However, he notes, the technology used in the switch could also be applied to other applications, such as IP edge routing, which typically uses shorter wavelengths like 1310 nm. Specifically, he says that routing vendors could use Lynx's protection switch to build line-card redundancy into their routers more cost effectively.
That's could, because Lynx hasn't yet managed to engage with any IP routing vendors. But analysts seem to think there will be plenty of interest once word gets out, provided that the big vendors haven't already found an alternative method.
Chris Nicoll, director of infrastructure analysis at market research firm, Current Analysis, says Lynx seems to have found "a problem worthy of solving." The issue of built-in redundancy in IP routing is huge, he says, and will likely become an even bigger issue as more service providers begin to deploy Internet Protocol (IP) in their networks. That's because most IP routers today do not offer the same kind of reliability and redundancy that’s available in other telecom gear, like telephony switches.
In the Internet core, providers have overcome this issue by deploying two routers in each location. Both routers are configured exactly the same. One carries live traffic, while the other sits idle. When a line-card outage or some other failure occurs, traffic switches over to the redundant router. But while this kind of protection is effective, it’s also very expensive -- too expensive for the edge of the network, where many more routers are needed and less space is available.
“If a service provider wants a completely protected router, he has to use half his line-card space for redundancy,” says Rafael Francis, senior product manager of Laurel Networks Inc., an edge routing startup. “While service providers like the redundancy, they hate the cost. They would like to have a 1 to N backup system where one line card is standing by for multiple line cards. But few vendors, if any, do that today.”
Francis says this is a problem that all routing vendors are looking to solve. Avici Systems Inc. (Nasdaq: AVCI; Frankfurt: BVC7) and Alcatel SA (NYSE: ALA; Paris: CGEP:PA) have already started building more redundancy into the architecture of their routers using high-speed electronics, says Current Analysis's Nicoll. Juniper Networks Inc. (Nasdaq: JNPR) and Cisco Systems Inc. (Nasdaq: CSCO) are also looking at this problem.
Lynx's Leigh says Lynx’s optical protection switch could provide the 1 to N protection sought by router vendors. It has designed modules to offer 1:4 or 1:8 redundancy, meaning that one redundant line card would sit idle to protect four or eight line cards. A Lynx module monitors the system and provides automatic switching to the spare line card whenever a failure occurs.
Laurel's Francis says this idea sounds promising, but he cautions that it would likely require a lot of integration with the router design to work effectively. Cost is also an issue. The price point must be low enough for IP routing vendors to consider using that solution instead of solutions they are currently developing, says Nicoll.
"Once they prove out the application, it could give next-generation IP routing vendors another solution to work with as they look for solutions to the redundancy and reliability issue," he says. "But I wouldn’t say there is a pent-up market waiting for a photonic switch on line cards. IP routing vendors are already developing other methods to solve the problem, and there is still the issue of redundant software and redundant control planes to be worked out."
— Marguerite Reardon, Senior Editor, Light Reading
www.lightreading.com Want to know more? The big cheeses of the optical networking industry will be discussing optical switching and edge routing at Lightspeed Europe. Check it out at Lightspeed Europe 02.