>>BUT, my DWDM question has to do more with the concept of the cost-effectiveness of increasing the capacity of the datastream vs increasing the number of wavelengths.<<
It boils down to the economies of scale, where all paths diverge and then must re-converge, once again, when it really isn't necessary in the first place in many situations.
If divergence is indeed a necessity, then that's a different story. As there will certainly continue to be enough network environments where SR is overkill, even counterproductive using current bandwidth packaging techniques, for a number of years to come.
I can temper this temperment, however, by pointing out the humongous leaps in capacities nor being, or soon to be achieved, with today's Avicis, Torrents etc., and DWDM itself, in comparison to the trickles that were achievable up until now.
In other words, I see no need to establish a glass ceiling (pun intended] at this time, which is my main point, and one of the main precepts of this thread, if I may be permitted to speak for my co-antagoists or protagonists [take your pick] here.
But if each SR stream is capable of significant multiple greater capacity, and if they could then be used to populate a number of wavelengths on a conventional DWDM, then I think that those efficiencies speak for themselves.
Recall that the premise here is that SR and DWDM are not entirely mutually exclusive to one another. The degree to which they can co-habitate and feed off of each other is a question that I think needs to be answered here.
>> "What if those 28 T-1s suddenly became comparable to T-3s in capacity without the corresponding increase in cost?"<<
I see your hypotheses here, but that's all it is, an hypothesis. Even if there is no apparent cost to splitting them, up up-front dollar-wise, due to massively dense integration techniques, there is an overhead of labor and administration involved nonetheless that is unnecessary in many instances. They are in those instances that I make my case.
[[ As an aside, your point is especially well taken in another venue, since in the SONET model, there are already 28 DS1s (T-1s) embedded, in a T3 in many cases where they needn't be. Here, T1s take the form of a virtual tributary [Vt-1.5] in SONET equivalent to ~1.7 Mb/s including SONET overhead.
There are, in fact, 28 of these Vt-1.5's stacked in an OC-1 or its electrical equivalent , the STS-1, that are managed on the fly, whether they need to be together or not. Touche! But, does this mean that it was necessarily an optimal thing to do? That's another story, too, and one, in fact, that would require multiple discussions in itself.]]
>>Now T-1s won't become T-3s, but what stops the current DWDM technology from carrying higher capacity datastreams in each channel? What are the difficulties? What are the costs vs increasing the complexity further with even more wavelengths? <<
I think that we are nearing agreement here. There are historical cost metrics involved with increasing bit rates in sovereign systems, with the max at this point in time being achieved at the OC-192 rate. That's roughly 10 Gb/s. Only one tenth the aggregate capacity that SR has claimed to be able to support in its prototype assembly. I don't want to trivialize the other complexities involved with engineering complete "systems" but this is the major one at this time.
Getting the densification of processing beyond the OC-192 rate on a single lambda [up until recently this could have been stated for a single "strand"] has not been seen as a viable alternative yet, due to poor yield results and enormously complex issues with regard to satisfying a number of other industry related norms. A box, in other words, must conform to its environment [which is far more extensive than merely talking to other similar boxes on a point to point basis], lest it be an isolated entity.
But lets say that they achieved a unified OC-768 rate on a single lambda. This is still only a small fraction of what SR has claimed to be able to support.
>>I just want to know what the DWDM industry might be able to do to easily counter this technology in the near future. After all they have the big advantage of being the currently entrenched technology.<<
I've heard other similar questions and remarks along these lines. I find it interesting that the industry should take up a defensive posture to this, instead of figuring ways of leveraging it. One means of leveraging it may result in rethinking or reintroducing an alternative optical spacing plan for DWDMs that would take into account the dimensions of SR's optical footprint. Anyone have any thoughts on this? This is one of the issues that Curtis B. so astutely brought up on another thread, although I'm not convinced that he was altogether right about this, and I was as unprepared to answer the question then as I am now. It's just a gut sense that I have, something intuitive, without being versed in the depths of their mathematical justifications, that multiple streams can be achieved. For this I defer to others. |
