x
DWDM

Ericsson Puts Its Own Spin on 100G

Ericsson AB (Nasdaq: ERIC) is trying out a 100Gbit/s technology that doesn't fit the Optical Internetworking Forum (OIF) framework but might have a place as a low-cost alternative.

Not yet ready in product form, Ericsson's 100Gbit/s technology uses 16-QAM modulation (meaning it encodes four bits at a time) that's been derived from the company's microwave and Long Term Evolution (LTE) work. Ericsson has put it through an 800km (497-mile) trial run in Sweden, as announced last week. (See Ericsson Assists in 100G Trial and Ericsson Completes 100G Trial.)

The point of the OIF framework -- a semi-formal standard for 100Gbit/s optical transmission based on coherent detection -- is to keep costs down by pointing vendors in the same direction. And Ericsson is working on an OIF-compliant 100Gbit/s technology, likely to be ready mid-2011, officials say. (See 100G Standards Aim for Lower Costs.)

So, why deviate from the pack with 16-QAM? For one thing, Ericsson sees potential in applying Ericsson's techniques to 400Gbit/s or faster speeds, a level at which the industry is still open to answers. (See The Terabit Ethernet Chase Begins and 100G Watch: Google Complains Again.)

More immediately, though, the technology might be just plain cheaper.

"It uses polarization multiplexing and coherent detection and DSPs [digital signal processors]. However, we have combined a simplified optical stage with a more advanced RF stage," says Tore Smedman, Ericsson's head of evolved transport platforms.

That last part means Ericsson can use cheaper chips in the optical stage. Specifically, the DSP gets its data at a slower bit rate and therefore doesn't have to be as sophisticated as the DSP in the OIF framework. (Whether that bit rate is low enough to use an off-the-shelf DSP, Smedman isn't saying.)

More of the processing does get pushed into the RF stage -- but that area has a ready supply of commoditized components, Smedman says. Again, that results in making things less expensive.

Ericsson isn't saying when (or even whether) 16-QAM 100Gbit/s will be put into a product.

— Craig Matsumoto, West Coast Editor, Light Reading

Page 1 / 3   >   >>
Sterling Perrin 12/5/2012 | 4:16:22 PM
re: Ericsson Puts Its Own Spin on 100G

Craig,


I'm not a big fan of deviating from the OIF approach for 100G, and I believe that most operators won't be either. The industry consensus around 100G has been a big plus for 100G versus 40G, and the OIF has played a big role in driving that consensus. At first blush, this alternative technology seems to be a distraction.


Ericsson may argue that there are some near-term opportunities while they (and others) bring out the "OIF approved" products. But it seems that investment diverted here takes away from investment that would have gone toward the end-game OIF 100G products otherwise.


Sterling

spc_markl 12/5/2012 | 4:16:21 PM
re: Ericsson Puts Its Own Spin on 100G

Craig,


It is very possible that Ericsson needs to generate a distraction.


Mark

spc_markl 12/5/2012 | 4:16:20 PM
re: Ericsson Puts Its Own Spin on 100G

Craig,


I am not saying that any particular type of distraction would necessarily do it any good.  However, a typical response from a large supplier that may be having problems is to divert attention.  I was also reacting to the comment made by Sterling -- of course, taking it to another level.  Otherwise, my discussions with customers in the past have indicated that Ericsson is certainly in the hunt regarding 100G.


Mark

Pete Baldwin 12/5/2012 | 4:16:20 PM
re: Ericsson Puts Its Own Spin on 100G

Mark -- Not sure I follow.  I don't blame you for being suspicious of a company's motives, obviously, but what kind of distraction do you think Ericsson could create?  The world already knows they don't have 100G now and that they're planning to have it next year, and there's no particular shame in either of those statements.  If they end up late with OIF 100G, I don't think the introduction of 16-QAM would make anybody forget about it.

paolo.franzoi 12/5/2012 | 4:16:20 PM
re: Ericsson Puts Its Own Spin on 100G

Yes it is a really bad idea for large companies to try out different technologies to see if there is a better way to do things.  Differentiation is very bad.


I mean seriously...if you read the article they were trying an experiment to figure out what was useful.  That seems like a completely reasonable activity for a large firm.


Will it turn out to be anything?  I have no idea.  Nobody does.  But at least someone is spending some time thinking about things differently.


seven


 

Sterling Perrin 12/5/2012 | 4:16:18 PM
re: Ericsson Puts Its Own Spin on 100G

7,


I view 40G development as the time when everyone experimented with technologies for higher speed transort and what we have at 100G is the consensus that emerged from all of those trials, successes, and mistakes. So for 100G, Ericsson's technology appears to be a bit of "reinventing the wheel."


If they're talking about beyond 100G (400G, etc.), that's completely wide open and fair game for vendor's claim leadership.


Sterling

paolo.franzoi 12/5/2012 | 4:16:17 PM
re: Ericsson Puts Its Own Spin on 100G

 


Sterling,


Experimentation never ends.  Different technologies scale at different rates and if someone thinks they have a better idea, it is completely legitimate to try it.  I assume that given your stance that Infinera should dissolve into a puff of smoke given that they attacked a solved problem.  See the point?


Conventional wisdom and standards are interesting guidelines.  If everyone followed them nobody would make any progress.  If E can show up with a part that is say 50% lower cost then the old standard will not matter.  If not, then it is an experiment that failed.  But at least they are experimenting and trying.  If not, then cut the R&D staff by 90% and let others do the thinking.


seven


 

Stevery 12/5/2012 | 4:16:17 PM
re: Ericsson Puts Its Own Spin on 100G

> I'm not a big fan of deviating from the OIF approach for 100G, and I believe that most operators won't be either.


1. The point is the suppliers have stopped caring that operators want a uniform interface.  


> and what we have at 100G is the consensus that emerged from all of those trials


2. It's certainly not true that 100G is the best technologies from 40G.  QAM is a tradeoff of S/N of the system for bitrate, so let's face it:  Honest optical guys would call this 4x28G, not 100G.


3. It IS about the business lesson:  If you let your customers demand a uniform OIF interface, then the product will not be produced at a profit.  The real lesson of 40G is that the companies who produced it successfully were either sold for peanuts or have sorry balance sheets.


4. We're finally seeing the result of the bubble burst:  Because there are fewer companies willing to produce high-speed interfaces, they (like Ericsson) are now dictating the terms.  What are the buyers going to do, not use the fastest interface and get beat by a competitor?  Supply and demand baby.

spc_markl 12/5/2012 | 4:16:16 PM
re: Ericsson Puts Its Own Spin on 100G

It seems like much of the discussion is taking place in an alternate universe in terms of some of the comments, which demonstrate a lack of intimacy with the 100G space -- or just do not make any sense.  


There is certainly nothing wrong with trying to differentiate a product and doing experiments.  However, it is going to be hard enough getting out the cost on serial 100G systems, even without deviating from the OIF standard.  Lacking a uniform standard, how the heck do you get sufficient volumes on components to get prices down?


Mark

paolo.franzoi 12/5/2012 | 4:16:16 PM
re: Ericsson Puts Its Own Spin on 100G

 


If you are using lots of existing parts that are already in volume.  That is what was described.  I have no idea if that would work, but it is an interesting idea.  I would not expect an analyst to understand that.  Analysis leads to a linear prediction of the future.  Vision predicts a disrupted few of the future.


And again, I ask the same question to you Mark.  Since it is clear that Infinera could not possibly have volume, there solution MUST be more expensive by your view.  I am not a big fan of their chances going forward, but they took a completely non-standard approach and ended up cheaper. 


seven

Page 1 / 3   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE