re: bandwidth supply and demand curves
===
The following is a question that was asked of me concerning the supply and demand curves that were shown in the link indicated, and my response. I posted it on the gtf forum last night.
===
[FAC]
I'm wondering what you think about the supply and demand curve that is towards the bottom of this link.
sciam.com.
Thanks.
=====
Hello Chris,
On the surface the chart shows some popular trending on the capacity curve, which, because of existing trajectories and some fairly reasonable assumptions concerning deployment rates, are easy to take a wild guess at. And that's all it is, that’s all it could be: a guess.
I'm not quite so certain, however, about the demand curve. Looking so far out, I mean. What the SciAm graph shows is what I call using some good old' Kentucky Windage. Looking out beyond 12 to 18 months can be a shaky call, at best.
About the only thing that you can be sure of is that bandwidth utilization is going to go up. I'm looking at several other projections that I have here, and two of them show similar trending, give or take five to ten Tb/s <cough!>
One of these projections by a fiber barron carrier (which was done using publicly available information in the press, and on the basis of past SEC filings.. (in fact, wherever they could get their hands on information without violating NDAs) that disclose funding plans. To paraphrase, its title is akin to: "Currently Funded Sources of Bandwidth Services and Supply."
The reasoning on their part is that they can determine - in a purely academic sense, and they did this I'm told in order to put the glut question into perspective - how long it will take each of a dozen carriers, starting with 360 and ending with Williams (well, maybe there'll only be 9 or 10 carriers, come to think of it ;) to fully light the fibers they currently have in the ground.
The basis for the algorithm used is the amount of capex that was needed to light the strands that have been lit to date. From my inferences, I believe that they also took other factors into consideration, such as improvements in technology, cost of capital, declining equipment costs per unit of capacity, etc.
Some really wild numbers fall off the page as you examine the report. In one case of one carrier whose name is a number (and whose days according to some are probably numbered, tool), they estimate the time it would take them to fully light their strands to be 39 years. A top tier IXC, 9 years. In the case of a recently merged IXC, 125 years <sic>. And in the case of a company whose stock symbol is a single letter above P and below R, 25 years. The mean number for the field was 38 years to light all of the fibers they currently have in inventory. Of course, this does not take into account any sales of those fibers that they might make to others under IRU terms.
Using this formula they estimated that by 2003 total capacity would be in the region of 95 Tb/s, assuming that all systems utilized 160 lambdas per strand by that time (quite a stretch if you ask me), and each lambda operated at 10 Gb/s. This would result in a yield of 1.6 Tb/s throughput per strand, in other words, which, in part, accounts for the technological advances that I referenced above.
Other projections that I’ve seen take into account straight-line growth, and then apply a polynomial s.w.a.g. factor to make it look like a hockey stick. I kid you not.
Assessing demand is a real hoot as far as I'm concerned. I consider net traffic to manifest patterns that those that are found in the study of climatological and meteorological occurrences. Chaotic. Look what happened in 1991 when Pinatubo blew.
It caused the same level of havoc in the environment on a global scale in '91 as the 'Net did in ’95-’96 to carrier and user infrastructures, not to mention the social and economic implications that ensued.
Both took place out of the blue, and both took the world entirely by surprise. And both were viewed as influences that stemmed from venues outside of the realms that they ultimately affected. Each of these was a classic case of what’s commonly referred to in some circles as, the “sucker punch.” And so it goes…
FAC |
