Couple of points here.
For the MX you have 4 queue's.
Realtime: if a packet is tagged with a PHB of EF (DSCP 46) then it will be placed in this queue and be transmitted immediately. You cannot configure traffic to go into this queue directly (you can only choose high, medium, low) but if the traffic is tagged with EF then it will be moved up into this queue.
High: This queue is a bandwith queue that is guaranteed a minimum of 4/7 of the WAN bandwidth, if the queue has bandwidth to spare it will yield this to the lower queue's.
Medium: This queue is also a bandwidth queue and is guaranteed 2/7 of the WAN bandwidth, this can also yield leftover to the low queue.
Low: This queue is the last queue with only 1/7 guarantee. It can of course receive spare bandwidth of higher queues.
For this to work you need to accurately set your WAN bandwidth values on the top of the configuration page.
On the MS switches it is different.
There you have the ability to set VLAN and port based rules that explicitly set a DSCP value or choose to trust the incoming DSCP value from the endpoint. There is also a link (dscp to cos queue map) which assigns a certain DSCP value to a queue.
MS switches alas don't have a priority queue but they have bandwidth queue's a bit like shared round robin. The switches have 8 queues where the highest two are used internally and the other 6 are configurable in values from 0 to 5. Each queue number has double the bandwidth of the queue below it. So you could for example choose to put AF41,42,43 traffic into queue 3, AF31,32,33 into queue 2, etc...
There are no advanced systems where you can re-mark traffic depending in some policing algorithm. So in essence you won't get the values 12,13,22,23,31,32,42,43 unless you specifically add them yourself.
