From the PR http://www.lightreading.com/do... I see that ADVA claims that BW efficiency is more than doubled compared to a 10x10G solution. So I assume that each 28G carrier goes on a dedicated ITU-T grid slot and ADVA requires 4 slots (50GHz spacing ??) for 4x28G versus 10 slots for 10x10G. This is a great advantage to 10G but still 4 times worse than 100G single carrier. I would like to see what cost difference can be achieved in reality, compared to a 100G coherent solution. If cost and power is better than 100G single carrier, there will be a good space for this technology in Datacenter applications I assume.
Sailboat, my intent was not to bash the ADVA solution. The referenced ADVA PR says "Spectral Efficiency - The 100G Metro solution outperforms the spectral efficiency of existing 10G solutions. Carriers and enterprises can more than double the capacity of their existing optical networks, thus protecting their network investment and extending the lifetime of their existing infrastructure."
So this does not make any sense in the context of your explanation, or am I missing something completeley. Spectral Efficiency should increase 10 times with your statement and ADVA PR would for sure put this right. 4 x50 GHZ = 200GHz versus 10x50GHz = 500GHz would be the correct math.
Spacing 4 x 28G in one 50GHz slot seems "magic", as this is for sure none coherent and filtering the 4 carriers so densely spaced is a real challenge.
If you can proof your claims, please do so. I am really interested how this is achieved and whether there is really some "magic" in this solution.
BTW: What kind of modulation is used per carrier, NRZ?
Nocopper, you missed the key point. ADVA is using only ONE ITU grid channel to do this. Simply they are doing some tight channel spacing using some kind of fine grain mux to pack 4 closely spaced wavelengths into one ITU grid channel that is totally transmissable on 50ghz ITU grid.
The ADVA guy in the youtube video claims less than 100W power consumption - this seems pretty incredible if it also includes optics. Does anyone know if the 100W includes optics or just the bare card?
10x SFP+ will burn ~10-15W. Assume the CFP is at the highest power class, say 25-30W. So optics alone will burn 35-45W. Hard to believe the rest of the electronics is only 50W.
Actually you are incorrect. CFP includes electronics in the module. The power for CFP is for both electronics and optics.
I am pretty sure the ADVA claim of less than 100watts for this card is correct. Including electronics and optics.
simple modulation scheme would enable this along with simple SERDES design. since there is no phase to manage, and no high power requirement on the modulation this seems possible. Remember, the big consumer of power is the driver electronics, SERDES, and other eletronics. not the optics themselves. I do not know what optics they are using nor what kind of electronics. But for short reach like ADVA addresses, and for simple modulation on each carrier, this is definitely doable.
First, these are not "my claims". I simply READ the press release which clearly states that the solution is putting 4 lambda's closely spaced into one 50Ghz ITU grid slot. Not my opinion nor claim. I have no need to "prove" anything to you. ADVA was clear in their release.
I do not know what modulation they are using. They have not announced it. But they did say it was not PM-DQPSK, but rather a more conventional modulation method that is lower cost for metro / enterprise / campus applications.
My read of this is that they are doing a classical "inverse" mux design for the transport which has been used many many times in communications networks. Given that electronics are either already native 28G, or are using "gearbox" designs to electrically up the speeds from internal to the pcb boards speeds of 10G up to 28G, having 4 28G streams should be relatively straight forward.
50G is the ITU grid spacing used, according to the press release. Read it again. Pretty clear. So it is NOT 4x50ghz ITU grid. you math is incorrect and indeed you completely missed the point of the release. They clearly state they are doing 4 carriers INSIDE a SINGLE 50Ghz ITU slot. the math would NOT be 4x50Ghz spectrum vs 10 x50g spectrum. Rather, it would be 4x28g in one 50Ghz ITU grid slot. As the press release states. ADVA did not provide math at how they achieve greater than 2x spectral efficiency as that is probably only shared under NDA. But I know the ADVA company and they are good at what they do. If you are a competitor, you would be wise to take them at their word. They are not known for over hyping. What they did say was more than double the CAPACITY of their own existing optical network platforms. They did not claim (see your own quote) double spectral efficiency. They claimed double capacity over their existing 10G solutions. The only stated that the 100G outperforms the spectral efficiency of 10g. They did not state by how much. you are confusing their statement on capacity in their platform with spectra efficiency.
NOTE re filtering: this is not magic as you claim. Do your research. tight spacing of DWDM filter technology has been around for a long time. First demonstrated by Ciena in the late 90's and also by Lucent. Then later by several start ups. Spacing as tight as 6.25Ghz was acheived long ago. I do not know who they get the filter technology from but this is definitely doable. ADVA does not make such press releases just for fun. they are well known to be careful and very competent at execution. My guess is that it is some kind of grating based tech for the filters, either AWG or perhaps Echelle grating, or maybe some kind of buik grating.
Not too hard to do, if you can overcome the thermal issues of center channel spacing over operating temp range. Clearly they have some solution.
This is the same thing Opvista did several years ago @ 10G and 40G and I'm sure Vello is working on this same thing for 100G.
Usable = Yes.
Innovative = No.