We'll answer this question with business-quality conferencing as our goal and focus here on using VT systems for two-way collaboration. That implies full-CIF video presentation, clear, non-distracting audio and good support of whiteboarding or application-sharing during a conference.

Studies have shown that after people accept that two-way visual telephony provides lesser broadcast quality than TV, they will tolerate variations in the rate at which video frames are presented-even occasional freeze-ups-without much complaint. Video is important for socialization at the beginning of a collaborative conference, but often is ignored when attention turns to whiteboarding and application sharing.

Audio is an entirely different matter. A telephone call is the paradigm. If the connection doesn't sound clear and distortion-free, or if there are noticeable delays (anything more than 100-ms is noticeable; 250 ms is annoying; 450 ms is unacceptable), most users will be discouraged. If you're managing a LAN-based system, you can mitigate this somewhat by explaining, "this will be similar to a telephone call carried over a satellite circuit; there'll be a noticeable delay between the time someone stops talking and you can start." Or, you can give conferees those big "We're Number One" foam hands-on-a-stick" to wave when they wish to interrupt a speaker.

Another strategy is to design your network so that one-way delays in audio stay well below 200 ms. Here is a strategic overview of QoS and bandwidth allocation on various transports: