Hi Eric,
Despite the size of the current dual-2.5 Gb/s backbone (the one that T's supposed to be putting in place at this time), it is not all that robust when you consider that other Tier One B-B providers are now quadrupling, in some cases increasing the ante by ten-, sometimes twenty- fold, through the use of DWDM-enabled routers which are also capable of terabit speeds.
Of course, T can match these speeds if they have downstream justification or incentive to do so. That is the question, though: Without last mile outlets to support the much higher speeds in the core, will they have a need to bolster the 5 Gb/s backbone anytime soon? More on the last mile, later. This goes geometrically for the sites that have multimedia to offer.
Secondly, as DownSouth has indicated, the robustness that does currently exist, or will exist when it's complete, on ATHM's mesh is isolated to a great extent, when it comes to pulling content from the majority of other content sites. They must peer with other top tiers in order to access it.
This means an increase is needed in the peering links and presence in a growing number of hand off points (exchanges) at commensurate levels of expense, as well... if they are to maintain momentum in accelerating their bandwidth supply to users.
Another possible glitch exists here. I touched upon it indirectly in my recent comment upstream about the 'net not being what it used to be. And that is, the nature and being of relationships between the 'net's major players.
As ATHM and RR reject or ignore the calls of other ISPs who want access to the local cable modem pipes, a form of retaliation may soon start to appear, en masse, at national access points (NAPs) and other Internet Exchange Points (IEXs) where peering arrangements are still very much a form of free and open market decisions.
I expect to see subtle forms of retaliation appearing there by those who have been scorned and by their allies, at first, if they haven't begun to contemplate these measures already, and are now only waiting to gage the eventual outcomes of some last ditch regulatory relief efforts, and the most recent acquisition bids.
They may eventually take the form of outright penalty assessments against those who refuse to honor the precepts that have made this all possible, from the outset. It'll all be another one of those growing phases, as the 'net undergoes adolescence.
The precept that is most lacking today, the one I'm referring to here, is that of peering with one eye closed, kind of, to the direct financial results of each component's specific contribution to the bottom line. The mass of the Internet is moving away from this at breakneck speeds, and there are bound to be some casualties along the way.
Geting back to my original observations again, and dismissing the more sinister possibilities in the more immediate preceding paragraphs, these all add up to some very expensive and technologically highly-ambitious initiatives. And more often than not, by the time that they are achieved, or even prior to completion, they tend to become anachronisms.
Caching comes into play, but caching does not address live, interactive stuff. And there is a limit as to how fresh, or relevant, cached material tends to be in a fast moving world. Having said that, caching is nonetheless an important ingredient in ATHM's strategy, I believe.
Then we have to look at the ATHM head end router nodes, i.e., the ones that are currently installed at MSO POPs. All but the newest of them are probably modeled after traffic profiles of two or three years ago, when email was still the most compelling reason to go on line. Do these routers have the port capacities to support massive flows from the Internet's core? Whatever that is, nowadays...
Then there is the local distribution. Even if the routers at the head ends could accept large flows from the core, where would they pass them along to, if the cable modem segments are crowded, which they will inevitably become with success, which is the strangest twist of irony of them all.
The congestion in the local segments will cause traffic to backup, anyway, despite whatever reinforcements are put in place at the router level in the H-E, or in its backbone capacity.
What you can deduce from this is that there will unlikely be an inevitable series of iterative processes at play, leaving room for interlopers to seek legacy weaknesses. There is probably no end is site as to how voracious end users appetites will be for bandwidth, once they get a taste of it, and what I'm saying should also signify that users wont be getting their fill of it for very long, as long as economic impediments exist along the way.
If the core backbone, along with the routers, are both bolstered adequately, the increases in bandwidth availability to the home due to the present incremental cable modem technologies, will only be perceived as increases until it is fully commercially successful. Somewhere near that point in time, it will be counterproductive, due to a run on the limited supplies of bandwidth in the last mile.
There needs to be more spectrum allocated to cable modem use, at first, and a stepped up move to using fiber closer to the home, if not into the home, over time. For the time being, cable modem users are enjoying superior performance in comparison to the techs that they were previously accustomed to. However, there is a finite limit to this opportunity to realize euphoria. All I can say to those who have it at this time is enjoy it while it lasts, but don't be surprised when those recently retracted use policies are reinstated. Users may actually be clamoring for it.
Regards, Frank Coluccio |
