I'd like to put some closure on the two posts [1860 and 1861] I wrote over the weekend, since one of my own senior consultants challenged me on some of the loose ends I left regarding the last sentence in post #1861:
"That is, until they [the carrier and end-user reference models] eventually collide with one another. Then what?
Here's how I replied:
[Greetings], of course, you're right. Knowing the context of the thread and Gilder's fiber sphere model in general would have been helpful.
I later characterized the two posts as being an odyssey of sorts which was flown at high altitude, indicating that metaphorically the issues I raised were correct, but not literally for the most part. Such would have required diving into some very deep waters, and that's not what the thread is about.
The collision which I mentioned? Of course, in introducing this kind of a dilemma, I merely highlighted the need for carriers and service providers to stay ahead of users where throughput speed and capacity was concerned. There was no intent to imply that at the OSI model, specifically, anything could be done, only that the NSP would have to upgrade their facilities.
Alternately, they would have to include the customer onto their network as a *peer,* i.e., let them on their network [provide them with suitable network access] using the same native protocols which were endemic to the carrier's own platforms, thus normalizing the dualistic appearances of two reference models, since each would view what was happening from the same relative perspective.
At that point they could conceivably be viewed as being in cadence with one another, and where the carrier was using WaRP or some other optical platform derivative, so would the end user, in kind.
In other words, at some point [when users' speeds and capacity expectations approached or exceeded those of the service provider's, limits] the customer would need to enter the SP's network at the same DWDM levels, or use the same next gen networking protocols [e.g., the SP's specific flavor of MPLS, etc.] that the carrier used, instead of the carrier being viewed as the Layer 1. In other words, whatever the Layer 2 or 3 protocols were that the carrier was using would be the same ones the user would use, at that point.
Since this is not something that users would find favorable or attractive, since it limits their own freedom of protocol choice, hence limiting their overall movements, I see this as being only marginally acceptable unless such a platform met with IETF sanctioned protocols, and of course that is not the case. Each of the current optical vendors is at this time assembling their own proprietary means to beat the traffic problem.
I therefore see the continued need for the carrier to appear as only the end user's physical layer (Layer 1) in those cases where the user is seeking a transparent core (no intermediate hops transiting the core), with users continuing to use their own selection of higher layer protocols. All of which means that the NSPs must continue to stay several steps ahead of their end users' (and subordinated carriers') utilization curves in order to keep the spoofing factor alive. Frank |
