To: Tulvio Durand who wrote (18500 ) 10/28/1998 6:49:00 PM From: Jorj X Mckie Read Replies (1) | Respond to of 77398
Hello Tulvio, Mind if I join the fray? The modem (or cable modem in your case) is only one factor in the connection. Even with a fast modem, I would guess that the local jitter (delay variation) is significant. And then if you go back to what the others were saying, it isn't IP that is the issue, it is the internet. The QoS that they are talking about is to guarantee that the voice packets are delivered in a timely manner. "Timely" is around 250 milliseconds. Anything over that and the voice quality will suffer regardless of the quality of the compression algorithm. Best case delay on the internet is about 600 milliseconds, and often much worse. So until the internet can reliably deliver the voice packets in a timely manner, there will be quality problems with internet telephony. The technology is there today, it's the infrastructure that needs to upgraded. One thing that Cisco is doing that is unique (maybe someone already covered this??), is translating the layer 3 QoS parameters to layer 2 QoS parameters (ok, IP QoS to ATM QoS). This allows the ATM backbone to recognize that IP "Voice" packets are different than IP "Data" packets and to further guarantee the timely delivery of the voice. As to free voice, I think that it was meant in a relative context....if an enterprise has a private network with both voice and data, you can make the carriage of the voice certainly a nominal relative cost. However, it is safe to say that no matter how good a feature is, if you are talking to a service provider and they can't charge for the service, you will not see that service. So, yes, service providers will charge for voice, but it will certainly be a nominal cost in their overall packet network. As to Lucent. They have the most to loose with the advent of multiservice networks and their strategy seems to indicate that they are trying to prolong the inevitable while still appearing to be moving toward supporting multiservice packet networks. Nortel, however seems to have accepted that packet networks will prevail, however there seems to be less confidence in their ability to execute over the long run. Tom