Network Engineering Stack Exchange is a question and answer site for network engineers. It only takes a minute to sign up.
Sign up to join this community
I am currently working with scheduling components that rely on the PCP information that can be found in the VLAN headers. I was wondering when and how these values are assigned to the frames.
How can priority groups be distinguished during priority assignment? Is all of this done during encapsulation?
when and how these values are assigned to the frames.
This operation always happens on ingress when traffic arrives on interface. The 'how' typically is a matter of a policy, which in arista/cisco world is a policy-map applied as service-policy on interface. The 'when' also depends on that policy, which could be various different reasons. There is also marking/re-marking that could happen on egress depending on thresholds and multi-rate policies (e.g. you can have one mark for traffic within certain bandwidth and then remark everything above with another CoS or DSCP value)
Priority groups is nothing but a logical separation. These groups can form queues, be attached to interfaces and follow different policies & queueing strategies. Most vendors have a concept of "priority lane" or "priority queue" which is taking the highest priority of the scheduler (basically, the scheduler will always serve all the traffic within the priority lane first, before it moves to the other lanes). This is very dangerous though as if it is misconfigured it may cause traffic starvation & thus causing issues to control & management plane protocols. Those queues are typically hard limited, meaning anything above the agreed rate will get dropped (or re-marked)
when and how these values are assigned to the frames.
There is no single way. Some devices tag their frames or packets by themselves (most IP phones), sometimes there's a policy on a switch, a router, or some other way.
You should be aware that not only 802.1Q-tag PCPs are in use but also (likely even predominantly) DSCPs in IP DiffServ.
Which one of the priority fields is actually in use, has priority, and defines exactly which kind of handling (number and size of queues, weighting, ...) depends on the installation at hand. There are "best practice" recommendations but they're not rigid.
