"If I have two seperate Priority llq classes and they are matched by access-lists,
How would they affect one antother?
They should both be prioritized, unless this is where the round robin comes in?"
My understanding, there's no round-robin between LLQ classes. All LLQ traffic is placed into one FIFO queue. If correct, one LLQ class can impact the performance of the other. However, since total LLQ bandwidth allocation is less than available bandwidth, LLQ is dequeued first, and LLQ is policed (per class) if there's congestion, impact likely will be minimal but there may also likely be some. (Actual impact would depend much on the attributes of the data flows, i.e. how bursty, and how the ratio of LLQ offered bandwidth compares to maximum available bandwidth [less to max should perform better].)
In other words, other than the LLQ policers dropping some class traffic, these two policies should provide like performance:
policy-map QoS
class VOICE
priority percent 15
set dscp ef
class VIDEO
priority percent 30
set dscp ef
policy-map QoS
class VOICE_n_VIDEO
priority percent 45
set dscp ef
"if so, then it would be better to change the polict to: "
In your new policy, you've no longer have video in a LLQ class. This will keep video from interleaving with the voice traffic, probably improving voice's performance, but at the same time, degrading video performance.
If the video is streaming video, this might be a fine policy. If the video is something like video conferencing, guaranteed performance might be inadequate.
