The real issue of concern is the performance needed to run voice either over IP or a traditional TDM. To meet the performance requirements to run voice over IP, the capacity of IP networks has exploded in the last few years so it's very easy to give the performance needed to support voice. The primary performance parameter identifying whether an IP network can support voice is jitter (variation in delay). To your point about being connection-oriented versus non-connection oriented (i.e. connectionless), there used to be concern about packets showing up out of sequence because they took different paths. Today, most IP networks have one very high bandwidth connection between major city pairs so a packet going from one location to another really only has one path as primary.
Related Q&A from Ted Studwell
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.