Ken, proponents of VoIP are stating that it will run on the third rail of the NY City Transit System. The ubiquity of IP, along with its robustness in hostile environments, are its principle strengths according to many, and the underlying carriage at Layer 2 is rendered inconsequential. Or, so the story goes.
Of course, the size and protocol of the conduit actually does play into VoIP's ultimate performance. As for which is "favored?" I think we begin to enter the realm of theology, at this point.
Most ITSPs who profess to have commercial grade voice at this time are electing to port their packets over private IP backbones. Of course, that's what they say, but the consumer is left in the dark as to the actual routes taken, and what their underlyings are. I don't think I could give you a conclusive answer to your question.
As long as there is enough headroom - that quality I like to refer to as invisible QoS - that could be affordably provisioned, it'll work just the same from a sound reproduction standpoint, with variations in latency and jitter, being the primary other parameters left to contend with.
I would prefer, in most instances that have to do with appreiciable distances, to see an underlying ATM structure at this time for time dependent applications like voice and video conferencing, due to its ability to support virtual hops in IP networks, which are routes nailed up in software, at Layer 2. As such, forwarding takes place at the speed of silicon switching, as opposed to table-driven software routines that could be extremely time consuming, especailly when older vintage routers are used in the all-Layer 3 model.
These are hard-mapped, as opposed to actual Layer 3 routed hops which depend on discovery and forwarding decision processes, at this time. The reasons for this stem from the times that are required for lookups along the way when using the latter (Layer 3) approach, entirely.
In ATM, on the other hand, these decision times are almost entirely eliminated permitting faster transiting times. Faster transport is always a primary concern in supporting a conventional conversational mode.
Next generation routers, it's hoped by many, will preclude these distinctions in the future. We'll see.
Reagards, Frank Coluccio |
