x
Optical components

Essex Demos 40-Gig Alternative

BALTIMORE – NFOEC – In what’s likely to be the final fling of this year’s NFOEC show, Essex Corp. (OTC: ESEX.OB) this morning demonstrated a thought-provoking example of what DWDM devices might (or might not) look like in the future (see Essex Hypes Hyperfine WDM).

The demo, in a hotel suite near Baltimore’s convention center, featured Essex’s “HyperFine” widget, a 16-channel multiplexer with 6.25 GHz channel spacing.

The channel spacing is pretty wild. 100 GHz spacing is more normal in today’s networks, and while plenty of vendors are talking about much tighter spacings, most of it is still under development. Ciena Corp. (Nasdaq: CIEN), for instance, has demonstrated 12.5 GHz spacing but doesn't expect to offer it commercially for a couple of years.

On the other hand, this demo comes nowhere close to what Essex was bragging about seven months ago, when it was claiming that it could cram as many as 4,000 channels into a single fiber, using 1 GHz spacing (see Essex Claims 4000-Channel DWDM). Give it another seven months and it might be talking about the same sort of spacing as Ciena, eh?

Essex says it isn’t like that. It’s just that it’s taken a long hard look at the potential benefits of tight channel spacings and concluded that 6.25 GHz spacings will offer the best opportunity for its technology to dovetail with existing DWDM equipment, according to Michael Piacenza, Essex's VP of business development.

Analysts, however, say it’s a classic case of a technology in search of a problem to solve. Essex “needs to find an application where high channel counts has a real benefit in the network," notes Mark Storm, optical networks program leader at Frost & Sullivan. "It's definitely thought provoking, because it affects how systems vendors architect something." Traditionally, tighter channel spacings have been deployed as a means of boosting overall system capacity. But things seemed to have reached an impasse on that front. In next-generation systems, tight channel spacings must be traded off against higher bit rates, leaving systems architects with the complicated question of which is the most cost-effective combination.

Essex is proposing that lots of slow channels will offer the best solution. It plans to use its HyperFine to multiplex 16 OC48 (2.5 GHz) channels together. This would deliver the equivalent of a single OC768 (40 Gbit/s) on the 100 GHz ITU grid.

Piacenza offers several reasons why he thinks carriers and systems vendors should consider this approach. For a start, slow channels travel farther without errors. "Fifty percent of fiber in the ground could not support OC768 data rates," he contends.

Second, he claims that it will simplify equipment at the edge of the network. Individual customers use bandwidth in tiny chunks, which are then aggregated before being moved to a switch, where they have to be unpacked. By using OC48 rather than OC192 or OC768, there is much less packing and unpacking to do at each node in the network.

Last, most carriers use OC48 extensively in their networks, so Essex's product would slot in without requiring them to throw away much of their existing investment.

Of course, it's early days. First, Essex has to convince everyone that its product works, and today's demonstration is the first step in that direction. The company says it is putting together four prototype devices, which it will deliver in August. One will go to a major DWDM components vendor, another to a test equipment vendor -- the HyperFine widget can be designed as a 50MHz filter, which would be great for making an accurate spectrum analyser -- and another to a systems integrator. The final prototype may end up with a long-haul carrier, but that hasn't been confirmed.

"I think everyone in the industry will scrutinize [HyperFine], to try and work out what they can do with it," says Storm.

— Pauline Rigby, Senior Editor, Light Reading
http://www.lightreading.com For more information on NFOEC, please visit the Light Reading NFOEC Site.

jmd 12/4/2012 | 8:06:16 PM
re: Essex Demos 40-Gig Alternative Can somebody explain the pros and cons of increasing channel counts verses increasing bit rates? The analyst (Mark Storm) implies that increasing channel counts brings diminishing returns GĒō the same point currently leveled against the alternative of 40G bit rates.
melao 12/4/2012 | 8:06:15 PM
re: Essex Demos 40-Gig Alternative "Can somebody explain the pros and cons of increasing channel counts verses increasing bit rates? The analyst (Mark Storm) implies that increasing channel counts brings diminishing returns GĒō the same point currently leveled against the alternative of 40G bit rates."

As i can see, when you have high channel counts the FWM can be really annoying, maybe putting too much loss on the system.

In the other hand, with 40G the attenuation is bigger and the dispersion is likely to be very problematic.

I really don't know which system would be the easiest to implement, using lower rates with high channel counts, or 40G with lesser channels.

Both systems sound very tough to be acomplished.

The strongest point in using more channels, is to use the current equipment that the carriers have.

cheers!
ivehadit 12/4/2012 | 8:06:14 PM
re: Essex Demos 40-Gig Alternative today seems to be a negative on 40G day at LR. the Essex article was primed by a negative article on 40G per "industry experts".

is LR just trying to appear smart or is this grounds for a conspiracy theory.
gpearson 12/4/2012 | 8:06:13 PM
re: Essex Demos 40-Gig Alternative The question is, why should one cram wavelengths so close if I'm not interested in thousands and thousands of wavelengths on a single strand of fiber?

It's probably a technology we'll all be looking at sometime down the road. But for now, most carriers would be happy to sell the capacity they already own.
pncguy 12/4/2012 | 8:06:00 PM
re: Essex Demos 40-Gig Alternative At last week's OAA conference in Stresa, Italy, the statement was made that Nx40G is cheaper than 4Nx10G. I cannot recall who made it, but they were pretty confident that they were right.

So, I side with the previous message writers: which is it? 40G is cheaper or 10G is cheaper?
mr_twinkie 12/4/2012 | 8:05:59 PM
re: Essex Demos 40-Gig Alternative Wouldn't a 6.25 GHz spacing bring about second order effects?

Would be very interested in their theoretical link specs for such a system.
mr_twinkie 12/4/2012 | 8:05:59 PM
re: Essex Demos 40-Gig Alternative "At last week's OAA conference in Stresa, Italy, the statement was made that Nx40G is cheaper than 4Nx10G. I cannot recall who made it, but they were pretty confident that they were right."

Probably trying to sell their stuff :)

I'd be curious to know what exactly was cheaper...the bandwidth/fiber or more importantly the actual cost of setting up an equivalent system in either 10G or 40G.

I would think (know very little data for 40g) that 40G would have smaller link budgets and ugly dispersion characteristics. Smaller link budgets equates to site relocations (carriers don't like this) and dispersion of course removes over half their fiber from the table. So there's a heavy fixed cost component to 40G as opposed to 10G.

But the carrying cost of one more 40G DWDM versus adding 4 more 10G would reduce over time, until at full capacity 40G is cheaper. The need for this multiterabit fiber now is obviously not apparent, and hence the lack of any systems out there.

I'm using the 10G vs 2.5G argument as a basis here, since it is essentially the same story with 40 vs 10 as it was 3 years ago with 10 vs 2.5.
calpole 12/4/2012 | 8:05:56 PM
re: Essex Demos 40-Gig Alternative n*1Gb/s has a good case for DACAS but not
in all optical domain. Transdmission cost
of n$1Gb/s may be less but switching and
OTU cost currently is pretty high and as you MEMs
is bit-rate transparent, so, with a
given MEMS economy is with a higher bitrate
side if transmission impairment is not an issue!!!
calpole 12/4/2012 | 8:05:55 PM
re: Essex Demos 40-Gig Alternative n*1Gb/s has a good case for DACAS but not
in all optical domain. Transdmission cost
of n$1Gb/s may be less but switching and
OTU cost currently is pretty high and as you MEMs
is bit-rate transparent, so, with a
given MEMS economy is with a higher bitrate
side if transmission impairment is not an issue!!!
optical_newbie 12/4/2012 | 8:05:52 PM
re: Essex Demos 40-Gig Alternative Does each wavelength of a DWDM system require a separate line card? If so, physical footprint would be a disadvantage to the Essex approach.
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE