The Holy Grail: Uniting Voice and Data

If there is one application that has been most closely associated with the word convergence it is IP telephony (IPT). Voice data integration in the same network has been a goal for many in the industry since the wall came down between voice and data in the late 1960s. It was in the late 1960s that packet switching emerged as a concept in the original ARPAnet. Packet-based data networks were designed with data in mind. Packets were moved through the network first-come-first-served. Switches could store packets before forwarding them, and would do so if the target output link was currently being used to transmit a packet. Packet networks emerged as queuing networks where bandwidth, delay, and jitter could all vary widely.
Circuit networks, the model on which traditional telephone networks were built, put a circuit in place through the network for a particular communication and fix the bandwidth, jitter, and delay of that path for the duration of the communication.
If these worlds were ever combined, circuits were borrowed from the circuit network to carry packets. Getting better-than-necessary behavior may be inefficient, but it doesn't break applications. Packets were never taken from data networks to carry interactive voice.
As time marched on, however, two trends came together. First, we created ways to add class of service (CoS) and quality of service (QoS) to the packet network. Second, the world of wireless telephony exploded, effectively lowering the expectations of the users concerning voice quality in the network. The IPT era began around 1996 and moved aggressively into the mainstream around 2005.

Daniel Casique EES seccion 1