Marco,
I didn't forget. I'd like to get to your comments on TeraBeam in a moment, but first I'd like to post an excerpt from an e-mail message which I sent out this morning. It briefly explains my position in general terms as they relate to last mile technologies, and networking developments in general. I stated:
-------------
"As a networkologist I have to be critical of all developments, not only as they appear on the surface now but also as I see their ability to scale over longer periods of time. Today's cable modem architectures almost always demonstrate (with some exceptions now from some of the earliest implementations which are already beginning to top out) splendid performance and will continue to do so until some earlier than projected up take point on the curve.
"There is about another one good year or so for most cable modem and dsl systems, in other words, before they begin to show signs of serious stress. Some may live out longer terms without the need for upgrades, and some shorter. [fac: late edit: This is a gross over-generalization, admittedly, but it gives you a good sense for where I'm coming from.]
"The more successful ones, along with those who have had time to grow subscribers for the longest time (namely some of the original systems which were deployed going back to '97-98), will show signs of exhaust, first.
"So, my criticisms are not so much aimed at how these [specific] CM and DSL systems perform now, but aimed instead at where on their trajectories I see them petering out on the uptake curve."
------------
Now, re: Terabeam, you say:
"I'm not convinced that their technology will be replacing fiber all that quickly either. It's one thing to prove in a small test environment that you can transmit laser beams point to multipoint, but it won't replace fiber for long hauls; if I were outfitting my corporate offices, I would take MFNX's fiber over Terabeam without a second thought about it. Not as many headaches, period."
I prefer to discuss this in terms of i-r systems, in generic terms, rather than just TeraBeam, since I believe that once the concept is accepted TB will be but a single player in this field, and I'm not convinced yet that they will be a dominant player. That remains to be seen.
I-R systems probably wont replace fiber very quickly, if at all, especially for mission critical applications, granted. What it will do, though, is introduce high speed access capabilities to segments of the marketplace which probably would never have had cause to increase line rates over the short term, or at all. And yes, it will also do some replacement or winning out over fiber solutions for outfits whose budgets dictate same.
But during the early chasm crossing stage, I see i-r "in the right situations" proving itself in as an enabling technology, fostering new applications in virgin situations as well as one that will "replace" other solutions. Although, it will also do a fair amount of replacement early on as well, aka cannibalization of existing services (be they dsl or fiber-based, or other wireless solutions) like I stated earlier, whether it is because of budgetary constraints on the parts of user organizations, or because of egregious bandwidth pricing disparities in the early going between i-r's straight line "cost of ownership" over a three to five year term, and the fiber carriers' willingness to meet this renewed challenge with more competitive pricing.
Long haul is not an issue here, it never was, nor is any serious short haul possibility beyond several thousand feet. Not unless we are talking about point to point systems such as LU's optic-air grade devices which could be implemented in a tandem configuration, as in multiple series-connected (in-line) links used to form a longer backbone route. Even here, however, one would need to examine a wide range of variables, each representing a certain degree of potential risk, and in many cases where fiber is available I see this as a no contest in favor of fiber when it is a mission critical decision.
Here, again, the i-r solution may very well prove in as a suitable backup facility for service restoration purposes, or in the case of cheap p-mp, a viable service for casual or redundant availability purposes. And again, where conditions are right, as a primary choice for price sensitive users, or where the shot is so ideal as to represent minimal risk.
If we stay with TeraBeam for another moment and address their p-mp model, specifically, there are a host of issues which I would like clarification on before commenting further on their longer term viability. These issues will be fleshed out during broader-based trials than those which have ensued to date in Seattle, and as the trades get their hands on the information, I'm sure. For this I can only wait, since there is where the proof will be in the pudding and not in what I can read in press releases or hear on pumped up cc's.
One earlier misgiving that I had, that was corrected for me over in the Last Mile thread by gpowell, I believe it was, was that of eye safety. From what I learned over there in LM, and elsewhere during the past two or three weeks, my information on this topic was both dated and irrelevant because of the wavelengths that TB chose to use. While I am still a little wary about this parameter (because the overall architecture will include more than the p-mp links from TB, such as the rooftop backbone i-r shots that may be provided by others such as LU), my concerns on this one point have been greatly reduced. But I remain a little wary until further information is available, especially as relates to the use of optical aids such as telescopes and binoculars, when looking into these devices (of all manufactures) without the appropriate filters.
Other areas of concern which I have at the present time, to name just two which have to do with their specific design tolerances, surround issues like alignment with the Sun during different periods of the day, and reflections off of neighboring windows from same, and issues surrounding how they manage upstream congestion from their hubs to the 'net's core, etc. vis a vis their stated throughput capabilities as a selling point.
Until those issues are answered, I'd continue to look towards already-proven systems for short haul point-to-point i-r operation, if I were in the market for extremely short haul links where fiber was not affordably available. For mission critical applications, though? I'd be inclined to agree with your closing assessment:
"I would take MFNX's fiber over Terabeam without a second thought about it. Not as many headaches, period."
Well said.
FAC |
