Hello Peter, good point. Glad to receive your post. I was beginning to think that the bizarreness of my highly speculative, albeit doable IMO, suggestion had scared everyone away. -g-
You asked,
"Why make 10G Ethernet a shared media when there is so much dark fiber? Would shared 10Ge be any less expensive than full-duplex 10Ge to install? to operate?"
I was obviously unclear about that part and should have elaborated some more on those points. At the present time there may appear to be a growing abundance of dark fiber along major routes, and in the backbones of a growing number of providers' networks, but the area we're talking about here is "the residence," if and once someone steps up to the plate to bring fiber directly to the side of the house, or into it.
And true, if this were dark fiber between the neighborhood field node and the individual home, then what you say is absolutely true. It would then truly be a dark fiber facility with all of its normally assigned bandwidth abundance that is usually associated with it.
But at the neighborhood field node, where I suspect DWDMs on chips will reside (perhaps dynamically tunable ones, at some point), we would see clusters of homes converging, each with their own spectrum allocations, whether they be static or dynamic (dynamically tunable). At that point only those allocated wavelengths for each home would be filtered at the dwdm level. So, we're really not talking about a full optical window, rather, we're looking at only a specific slice of the spectrum. Granted, this slice could be 50 to 100 GHz in width, but still, this would only be a "slice" of the larger 25+ THz available on the full backbone strand.
[I suppose one could ask, or argue, "Why make it opaque by introducing Ethernet protocols if you've taken fiber this far?" Anyone care to examine that aspect?]
I suppose that in the back of my mind I suggested to start off with shared as an expedient, only, since going switched, and then to the IP layer would add complexities that would only serve to delay matters further.
When demand begins to really escalate, after some period of shared, individual home segments, and then perhaps subnets within the home, could then be established through Layer3 switched alternatives, perhaps using some established vLAN techniques.
This is all very stream of consciousness, taking a wild stab after some previous brainstorming with some of the folks here earlier. I've not even begun to really figure out how an end user interface would fit into this scheme, or what kinds of protocols would be used to effect regular program video selection, much less digital video interactive services, but I have some nascent ideas.
If switched were just as affordable as shared at the silicon level, however, I would tend to agree with your drift that it should be used right out of the gate. Comments and corrections welcome.
Regards, Frank Coluccio |
