Your rant was interesting. Did you take it far enough?
I suppose one could argue that the trick is to oversupply bandwidth by some measure in order to create enough "head room" so that applications don't hit their heads on the ceiling, and in so doing keep ahead of the actual utilization curve. But to do this correctly would require the kind of predictability and dynamic resource allocation scheme that could only exist through Divine Intervention, apparently. I say this because I've yet to see anyone do this both successfully and consistently, in over thirty some odd years of slinging dots, dashes, ones, zeros, and various other bits of information.
The introduction of fiber, and then dwdm later on, has not brought us out of this dilemma, it's only pushed us deeper into it while at the same time raising the stakes with each additional megabit made available. While there may appear to be a momentary reprieve in some venues, where bandwidth was once scarce, and where obtaining said bandwidth is concerned, IMO we will soon see demand outstip this bandwidth once again, if not in the absolute amounts of bandwidth which is needed, then in the ability to orchestrate it. You made that point very clear, as well.
And that includes the largest domestic and international carriers, the world's most affluent financial institutions, and the military. Of course, there are economic reasons, as well as supply side constraint reasons, why they'd like to keep it this way.. or at least there were reasons for controlling supply until recently, but even those have yielded to the new awareness of "the optical economy."
But still, the problem persists, and although we may have more bandwidth at our immediate disposal today, the next steps up the spectrum ladders will be at a much steeper slope than the last.
==
Prior to the advent of dsl and cablemodem (which we now take for granted, but will both be obsolete before you can shake a strand of fiber at them), I remember the machinations that industry pundits went through in order to get it from 1200 to 2400 to 4800, and then up to 7200 bps, and then 9600 bits per second, in defiance of some well established laws and names who had convinced everyone that it couldn't be done. Nyquist and Shannon, to name just two.
It took ten years, in fact, to get from 1200 baud in the late Sixties to 9600 bps in the mid Seventies. And at the time, the top speeds being derived from T1 sytems (which are themselves predicated on a digital 1.544 Mb/s system speed) was only 200,000 bits per second [200 kb/s] through oversampling of analog waveforms that were force fed through channel banks. I kid you not. It didn't matter, though. I/O ports on conventional data terminal equipment couldn't handle anythng faster, anyway. For, 200 kb/s was enough to overwhelm a mainframe back then.
At that time we would have thought of anything approaching today's dsl speeds over copper, or cable modem speeds over coax, as being near infinity.. or at least in that same vein.
In contrast, today we speak in terms of a potential combined 150,000,000 domestic cablemodem and dsl lines all operating at once (always on, always sending agents, always doing surveillance, always accommodating surfing, always, always, potentially, generating yottabits of data for residential use alone, in addition to the yotta yottabits [see: studwww.rug.ac.be ] that will be flowing in B2B and over enterprise intranets.
It's not the bandwidth. It's the ports. It's not the ports alone. It's adiminstration. And a lot more. Bandwidth might well be free. Administering it isn't. |
