Hey Nhan,
Great that you've checked out the documents prior. It's quite a big topic and challenging to take a simplistic view or use a standard calculation/template. The recommended practice on LAN's for TP -> Voice, Video in PQ - to guarantee the TP experience.
Essentially, the SRND's concepts are formulated as guide or starting point based on a defined tested environment (such as throughput, QoS Feature support, appropriate buffers etc) then engineers should adjust according to their constraints and priorities. You could approach through theoretical modelling or real world testing (Proof of concepts)
-> is it possible for voice traffic to be delayed if voice and video are in the PQ, from the perspective of a 3750
------> Generally, issues occur when there is contention for bandwidth. If you have enough bandwidth and no latency issues introduced by the transport medium (such as WAN links) -> then no issues.
Simplistically, You can also think of it as - TP 30fps = 33ms a frame. A frame can be 1 or more packets. There are bursts for interactive video but there'll be other mechanisms at play at the encoding/decoding/jitter buffer etc
RTP voice - steady stream of packets 20ms apart
So there's room for both
-> If TP traffic is outside PQ, does it cause pixilation? -> again, if there is contention for bandwidth then yes. It's a matter of allocating enough. The practice of placing TP outside of PQ is typically used where real-time class circuits are expensive.
HTH
Eugene
