gp -
I think that if the attribute was actually "demand" and not existing "bandwidth", then perhaps Home was correct, as in: (from your last reply to me:)
"... Given ATHM's recent statements about bandwidth demands being more symmetrical than envisioned, at some point the upstream constraints will be addressed."
Is that what they claimed, demand, instead of bandwidth? I can't find the reference. If so, I agree.. they must do something to de-couple or otherwise reduce the sizes of the large clusters, in that case.
"However, I don't see how it will be done within the current DOCSIS specification and with the equipment currently deployed."
I've maintained that same belief for some time, as you probably know.. but we don't have to belabor that one just now. Anyway, my tendency now is to soften up a bit, but not so much so as to lose my orientation. I say 'soften' because I'm seeing a gradual movement --as though a just-in-time regimen of provisioning were the best alternative to meeting the tsunami we've come to know as the Internet-- to performing upgrades over time, one way or the other. "It'll get done, don't worry," is the sense I get.
Especially if there is competition to force the issue, and especially in upscale and high-utilization locales with high exposure.
The problem with this approach is that it calls for multiple, time-consuming and opportunity-depriving, upgrades over time, which tends to aggregate costs in a way that could best be described as Gestalt in reverse.
In other words, the arguments for micro-incrementalization at one half the pace of Moore's Law fall apart when viewed in an historical perspective five years from now. Maybe sooner. This has to do with delaying an enabling infrastructure for the sake of preserving a minimalist infrastructure, which is a carry-over form of thinking from the days when the operators were rightfully assessing risk, when they were wondering if this 'Net thing was for real, or not. Well, it's for real, and someone had best tell the operators about it before penetration rates reach the levels they want them to.
----------
The long, drawn-out multiple upgrade approach stands in contrast to doing "the least common denominator" once, in a way where the basic structure of the outside plant doesn't change over time. Some may argue that the backbone in HFC has achieved this already, and that's true.. but it didn't go far enough.
In other words, the skeleton, right up to the home, has to be put in so that it will last at least five or six years, if not ten or more, instead of it undergoing surgery every two or three years, while unnecessarily holding back (and in the case of return path, resisting) the commodity we call bandwidth from paying subscribers.
[[Begin Sidebar: How many times does one imagine MFNX will be fibering up 40 Wall Street for The Donald ? They have one sleeve going into the building from Wall, the other coming in from around the corner. They lay an ample amount of high-grade singlemode fiber, once. What they hang on the ends of those strands may change over the next ten to fifteen years, but the basic substance of the underlying media will remain constant for a long time. End Sidebar]]
----------
Subscribers have become accustomed to the idea of having bandwidth at their fingertips, and they forget that it was only a year ago, maybe two (and many are still doomed to this form of purgatory, indefinitely), that 56k Dialup Modem was the highest speed that over 90% of the online population could hope for, or afford. Or, that could be supported by their ISP.
When you draw the utilization curve of today's "broadband" distribution networks, however, and look at the trajectories using the past six years as a ramp up, you suddenly get the see visions of a brick wall that can't be too far into the distance. Doubly so, when you consider the upper limits of most systems' capacity and the potential demands that can be placed on them, and taking into account that service penetrations are barely beyond their nascencies.
But the piŠce de r‚sistance to these predictions lie with the eggheads and appliance shops. They have been fast at work over the past two years readying a plethora of bandwidth chomping applications and devices which are now approaching their time for deployment. Whether they will be permitted onto the @Home (and other) closed intranet(s) is another question, altogether. Here, the acceptable use policies come into play again. Will there be curfews, measured usage, prohibitions? iDunno. Measured usage stands out as a distinct possibility, though, especially if these m-m and other streaming services are branded by the owner of the intranet. In this case, that means $Home.
But we don't have to look to far for "developers'" inputs, here. Home itself appears ready now to enter into alliances with webcasting ventures, themselves, which I find rather interesting because it will force the hand of their owners to ensure that whatever they intend putting onto user pipes can be supported. I'm still deliberating over this one, though, and don't know just what to make of it. Sure, it's rather safe to deploy webcasting at this time.. witness Franz's enjoyable experiences over the weekend, as he recounted for us earlier today. But keep in mind the still-early stage that the network finds itself in at this time, and what the consequences might be as the take goes up.
It's for these reasons, combined, that I envisage a need for some expediency at this time, even a sense of exigency --consider, it takes one to years for a large system to undertake a re-blocking upgrade-- on the parts of the MSOs if they are going to stay ahead of the demand curve.
--------
Of course, some of them think that things are just fine now the way they are, and some of them will be content to just play chicken with the ILEC, to see who could stay marginally better than the other in a perverse form of mutually-beneficial, cost-saving way. Re this last statement, see the following excerpt from a post I did on the LMT, recently:
Message 13805603
Referring to the ILECs and MSOs,
"... what they will do, instead, is play chicken with each other. By this I mean they will each look at the other's performance and do a kind of arbitrage on how bad they can remain without suffering any consequences (such as subscriber churn).
"This, come to think of it, is what they've done all along with deployments in general. Is it a coincidence that in my neighborhood I suddenly have both dsl and cablemodem to choose from, when four or five months ago I had neither (from an incumbent)? [6/6 edit: And only last year there were no solid projections by either that I could plan on, at all?]
"This is easy to do. If you are an ILEC, say, you install sufficient extra bandwidth, temporarily, to ensure that the platform kicks off properly. Then as the competition's performance begins to degrade because they've been around for a while and they're beginning to increase their contention, then you begin to throttle back to the point where you are just a tad better than them, but not by much. Or, you simply don't add additional bandwidth when you would have...
----------
Instead of the MSOs working their way through another four or five upgrades until they realize that they've reached the theoretical limits of coax and RF schemes over same, I think that some of them should be considering a more optimal base technology now, and simply bite the bullet. In the end they will have saved money, but more importantly, they would have enabled a host of far more dynamic applications than they are willing to permit (or could support) at this time, and create additional revenue generating opportunities.
HDTV over regular RF-based cable is a thorn, for the bandwidth that it consumes and the ad displacement factor which it creates when multiple less demanding commercial channels cannot be aired in its place.
But HDTV over IP or ATM? On demand, yet? Think of the possibilities. Can't do it? Sure, that's correct, it can't be done if no one puts in the enabling infrastructure.
----
I'll let you know if I come across any references to modified upstream spectrum plans. It's been good chatting with you.
BTW, this is the second time in several months that you've honored me by way of correction. I'm still indebted to you for catching me, early on, on the matter of eye safety icw infrared free space devices.
To refresh your memory, we were discussing older models of I-R devices whose standard wavelengths were in the ~600 to ~850 range, as opposed to the newer ones which use ~1500. Thanks again.
FAC
ps - this message was also sposted to FCTF at:
Message 13840590 |
