LX4 Gets Another Chance

It's ironic: So many components companies are waiting for their markets to come back... and yet, a market that is coming back has drawn shrugs from optical transceiver vendors.

A 10-Gbit/s Ethernet standard called 10GBASE-LX4 is drawing renewed interest from the likes of Cisco Systems Inc. (Nasdaq: CSCO), which plans to ship LX4 modules as early as the first quarter of 2004.

Most vendors, meanwhile, gave up on LX4 or never even tried it; and there's reportedly an alternative format that's being prepared for consideration by the Institute of Electrical and Electronics Engineers Inc. (IEEE).

Cisco, being Cisco, is ready to forge ahead anyway. Bruce Tolley, Cisco senior manager of engineering technology, says the company has "three or four" transceiver vendors lined up as suppliers, although he wouldn't name names.

LX4 is intended to send 10-Gbit/s Ethernet across 300 meters of old multimode fiber -- in other words, it targets the installed base. "10-Gig on that kind of multimode fiber goes about 20 meters," says Bill Woodruff, vice president of marketing at BitBlitz Communications Inc.

LX4 has long been included in 10-Gbit/s Ethernet standards talks, but interest among vendors waned during the dotcom boom, as huge installations of singlemode fiber seemed imminent.

Of course, the boom ended. And, as 10-Gbit/s Ethernet begins its ramp-up, it's looking as if the most promising customers will have to run it over old cable. Suddenly, LX4 is interesting again.

"We expect half the 10-Gig market is going to be for multimode -- that's in risers and enterprise accounts," Tolley says.

Tolley's been pushing LX4 since at least 2001, so Cisco's interest is no secret. Still, LX4 is a bit weird -- it avoids some problems of multimode fiber by using four wavelengths in a kind of coarse WDM. As the 10-Gbit/s Ethernet standard jelled, only Blaze Network Products Inc. and Molex Inc. (Nasdaq: MOLX/MOLXA) seemed interested in announcing LX4 transceivers (see Molex Makes Muxlink).

Some, such as Picolight Inc., shunned LX4 entirely. "We have felt the economies of scale favored serial optics," says Warner Andrews, vice president of marketing at Picolight.

"Certainly you hear a call for the application, but I don't think people believe LX4 is going to make it in the market. There are too many people who have backed out," says Steve Joiner, director of marketing for transceiver maker Ignis Optics. "Bruce [Tolley] would tell you emphatically they need a solution for this space, but I'm not sure he would mind if it was LX4 or something else -- just get it out there."

"Something else" might be arriving soon. Reportedly, there's a proposal afoot to replace LX4 with a serial option that uses electronic dispersion compensation to overcome the handicaps of multimode fiber. The IEEE 802.3ae task force, which shepherds the 10-Gbit/s Ethernet standard, is expected to hear the proposal at its September meeting.

Still, there's no guarantee that the new proposal will work out, leaving transceiver vendors with few options for supporting older fiber. "If LX4 turns out to be the best way to do that, then we'll have to find a way to do that," says Picolight's Andrews.

— Craig Matsumoto, Senior Editor, Light Reading

Standardsman 12/4/2012 | 11:35:06 PM
re: LX4 Gets Another Chance
I know that Molex are pushing LX4 but I thought Blaze had gone out of business.
Who are the other couple of vendors Bruce Tolley was talking about?

I wouldn't bet on LX4. There are a lot more 10GBASE-SR transceivers on the market now and with 10GBASE-CX4 coming next year for the Data Center, it will be a crowded space.
bwoodruff 12/4/2012 | 11:35:05 PM
re: LX4 Gets Another Chance The dynamic here is interesting. I'd love to see the LX-4 advocates talk more to why this may have legs. My company is taking care of the retimer chip needed to knock out a LX-4 module for the Xenpak type space.
Meister 12/4/2012 | 11:34:57 PM
re: LX4 Gets Another Chance "Have legs"? You must be kidding! This dog won't hunt - & it never had a chance. Just ask Agilent and all the other credible vendors who passed throughout the years. LX4 was granted a reprieve with the down-turn and even then it couldn't be made to work.

Long live serial ...
Pierre 12/4/2012 | 11:34:46 PM
re: LX4 Gets Another Chance All,

Here's an interesting development: while we have been assuming that Hyperchip had gone into hibernation, and was slowing buring through the remaining cash balances, they have recently posted job openings.

While development/R&D positions could be understood, even in this hibernation state (turnover), non-core hirings such as their business analyst posting makes me ask: what gives?

While quite broad, some of the responsibilities include "demand/capacity planning and sales forecasting". Are we nearing some announcements regarding actual product sales and/or a major roll-out?

Looking forward to your thoughts.

bradbooth 12/4/2012 | 11:33:43 PM
re: LX4 Gets Another Chance This was pretty much still-born when Agilent pushed so hard for LX4 and then changed their minds once it was adopted into the baseline draft. Although LX4 is going the way of the dodo, it may have influence on future Ethernet developments much in the same way 100BASE-T2 influenced 1000BASE-T. Right now, it looks like electronic dispersion compensation (EDC) is finding favor as a way and means to replace LX4.
lightnash 12/4/2012 | 11:33:34 PM
re: LX4 Gets Another Chance Civcom who just bought XLight are also active in LX4. They have still not announced anything in LX4 but they are working on this right now.
pigfaquer 12/4/2012 | 11:32:35 PM
re: LX4 Gets Another Chance The whole point behind LX4 is that it is the only 10G technology that doesn't require the IT Manager to lay new fiber. This isn't a problem in a 50 person office, but in a larger enterprise network, it would be a significant task to install new fiber.

You may have heard that 10G serial 850nm can reach 85 meters. Well, that is new MMF 50um, not common 62.5um FDDI grade MMF. 10G Serial can only do 26m over this grade which is what you would commonly find in an enterprise network.
bmullooly 12/4/2012 | 11:32:34 PM
re: LX4 Gets Another Chance >You may have heard that 10G serial 850nm can
>reach 85 meters. Well, that is new MMF 50um,
>not common 62.5um FDDI grade MMF. 10G Serial
>can only do 26m over this grade which is what
>you would commonly find in an enterprise network

Valid point. The MMF has to be OM-3, there's not much of this installed, so you need to pull new cable, so why not pull single-mode?

Only reason for LX4 in short haul would be if it's a lot cheaper than real 10G LX..

James_x_zhou 12/4/2012 | 11:25:41 PM
re: LX4 Gets Another Chance I am wondering the principle and structure of LX4, especially for the unusual requirement on each optical component:
i) Physical Medium: Fiber type
ii) Tranceiver and Reciever
iii) Route or Wavelength Management
iv) Power Management

If some web link can be available, pls kinldy let me know. My email is [email protected], TIA.
OptixCal 12/4/2012 | 11:19:34 PM
re: LX4 Gets Another Chance This may sound like an odd question, but does any of the LX4 stuff have anything relevance to anything @ 1550nm, or is LX4 strictly and specifically focused around legacy (old) fiber?
Sign In