Peter, Ray,
Briefly:
I think that there may be a point of confusion here stemming from the following: It's presumed, probably incorrectly, that the VoIP call would be carried over the Public Internet. Instead, the VoIP portion of the call would either end at the HE, at which point it takes a PSTN format to the remote end (the called party); Or, it might hop onto a private IP or softswitch provider's cloud, which I'm now told by folks I trust, demonstrates an equivalent level of performance to that of the regular PSTN connection. Of course, the latter claims can only be supported on the basis of the individual carrier that is used and how they provision/balance their cloud.
Expanded discussion:
There is no singularly defined method of deploying VoIP in MSOs' or other types of access venues. The basic VoIP solution, as it would be deployed by MSOs (at least initially), lies between the user's premises and the MSO's POTS component within their head end. OR, between the user's premises and the appropriate carrier POP after being routed or otherwise passed through the head end to the designated carrier's POP.
Under most situations, that's where the IP part ends and the regular PSTN DS0 voice channel begins. There is also the possibility that the MSO would hand off to a softswitch carrier, which I'll address in a moment.
The VoIP gateway that I referred to above could easily be a shelf in an ILEC tandem switching location, e.g., in which an integrated router/gateway is housed. Or, it could be a T1 or T3 connection to a softswitch provider's IP backbone, such as L3's or a growing number of others'.
Keep in mind, however, that such a softswitch and IP backbone arrangement is a special IP overlay arrangement, and not part of the open Internet, per se. It's the equivalent of a private IP backbone, in other words.
But if we ignore the softswitch carriers offerings for a moment, from the head end out to the WAN it will likely be a gateway connection onto the "regular" PSTN until some future point in time. So, for the larger part of residential end users needs, that's where the voip format ends, at the HE or first carrier POP (where the MSO might be colocated). It's an isolated solution in its basic form.
Of course, if the MSO wanted to further extend the VoIP connection over the WAN they could use a private IP backbone resource. Or, in the case of CableLabs packet voice initiative <not sure of the name of this initiative at this point, is it PacketVoice?>, there would be resources reserved between the cooperating MSOs for such purposes, possibly via a private IP backbone between them, or something similar. But even here, when the voice call had to leave this quasi-private cloud, it would have to take a gateway connection out to the PSTN, if it isn't already doing so via some other means (such as being part of a softswitch carrier's cloud).
But I doubt that the MSOs would attempt to pass the Internet proper off - leaving users exposed to an unpredictable number of hops - as a commercial grade of POTS - at this time. Or ever.
In the case of the gateway solution, what is the benefit of such an isolated point solution [between the end user and the gateway]? It's my belief that the primary desired result would be the conservation of precious upstream [and downstream] bandwidth on the coaxial section, for one, while at the same time posiitioning the MSO for future upgrades in the IP space.
This, as opposed to the more "lavish" circuit-switched alternatives using G.711/ds-zero channel formats, thus consuming far more bandwidth. And this would have the secondary - possibly more profound - effect of limiting the number of users that would be allowed onto the system [a critical design constraint] at any one time.
Comments and corrections are welcome. I think I'd like to hear from Denver Techie on this one.
FAC |
