Eric, I don't think that you are uniformed, nor do you miss much, and I don't think that you give yourself as much credit as you deserve. If you are "missing" anything at all, then it's the heuristics that one obtains after getting one's butt kicked for thirty plus years when you run out of bandwidth every time you go to install the stuff, before it even gets fully provisioned. The rest of this post approaches technical subject matter in a reader-friendly way. Anyone not inclined to read critical assessments and such, please press next.
---------
Within your post you provide the answers to your own questions. By citing that the DSLs and Cable Modems have taken such a long time, how many years? to get rolling? is ample reason why the fttc/ftth initiatives need to be commenced now, and not in five years from now.
Yes, it will take a long time --especially if left to be completed by the existing powers that be-- but if the DSLs, CM providers and the industry in general waits until they are all tapped out, then there will be an even greater penalty to bear by all users at that time, for obvious reasons. By that time, they will have become accustomed to a period of high speed access, and will be left with no recourse when the wall is hit. Providers need to begin installing deeper fiber, now.
Another set of items which you may be missing are the subtle forms which "utilization creep" take, until a number of performance dynamics are aligned, and then WHAMO! It's avalanche time.
As penetration rates increase to 20% to 25%, and utilization on the system approaches, say, between 30% and 40% during peak periods, things will continue to appear just fine, and in fact they will be.
Let's suppose that further penetration subsides, and for the purposes of this discussion we stay with 25% for a time.
Gradually, those limited number of users begin to use more demanding forms of applications both in the downstream (such as Solid's post demonstrated this evening) and in the upstream (where they are not restricted by the provider due to the paltry state of affairs in the upstream direction in HFCs).
Before you know it, due to the higher demands placed on the system by these new applications, the same number of users who were stressing the system at only 40% utilization are now stressing it to 80% or 90%. sometimes topping out at 100%. Still no major problems yet, but some signs of stress are beginning to appear, even if only sporadically.
But of course, the penetration rates don't subside when user activity increases. Just the opposite occurs. With increased utilization and added functionality afforded by the more demanding and feature rich applications, Metcalfe's Law begins to kick in, which means simply that usership always tends to increase in these situations, not decrease.
This always has a tendency to spawn a vicious upward utilization effect when enough "head room" is not built into the design, but things are still viable here, for most of the time, but now there are brownout periods that begin to last for a half hour or more, and sometimes prime time is unbearable altogether.
These new applications nonetheless gain in popularity, and before you know it you are simultaneously doubling your number of end users on the system, driving utilization now to 120% to 150%. Since 120% to 150& is an impossibility, of course, you begin to back down on individual user speeds in order to compensate, or new curfew rules and additional restrictive policies are put into effect in order to further stem the tide of traffic.
All this time, other applications that were in development in order to 'finally' exploit the promise of broadband are suddenly brought to market, and their uptake begins to add to the burdens, further. This will raise utilization demands up to the 200% to 250% mark, and above, which means that "actual" speeds per user must throttle down to one-fifth to one-tenth of the system's rated speeds, if the system doesn't melt down, entirely. Note: these are not linear relationships; once you go over a prescribed bogie on a contention based, or quasi Ethernet based system, shutdowns come into effect to the detriment of all.
Add home movies being sent over the 'net to the two-way mix during holiday periods, and you've all but brought the system down to its knees.
Okay, so I'm talking two to three years out, optimistically, for the most part here. But remember, if you wait until that time to put in the deeper fiber alternatives, then you will be at that future point in time right back to where you were two to five years ago, futzing around with V.34 and then V.90 modems, while aspiring to be one of the first on the block to be a part of the new optical platform when it comes out... three years hence from that time.
A big mistake that many folks make is assigning a linear relationship to the number of users --as they increase-- to the amount of additional bandwidth demand that will result. From the CED article that Solid posted tonight it should be clear that new users who come onto the system in six months from now will immediately, at that time, be using at least five-hundred to a thousand times more bandwidth than users who were on the system only six months ago looking up definitions on the meriam-webster dictionary site. Multimedia will do that to you, in a wink.
Regards, Frank |
