How QoS policies work
In a QoS policy, packets are classified and queued. QoS policies are applied to a stream of packets after forwarding lookup.
Classification is the process of splitting up packets into separate streams based on some criteria. These criteria are a subset of the packet classification that is used in policy-based routing and firewall.
QoS classification allows packets to be matched, based on the source and destination values of IP and MAC addresses as well as DSCP and PCP values.
On the software platforms, classification is a two stage process. On the hardware platforms, only the second stage of classification is done.
- This first stage can split the traffic into multiple classes. You can associate a different QoS profile with each class.
- The second stage of classification uses the packet's DSCP or PCP value to split the class's stream of traffic down even further into sub-streams that are directed at individual queues.
class) is different to a traffic class (
The QoS classification process assigns a packet to a class. These classes are identified by one or more match rules.
Classes are evaluated in numerical order. The first class that matches is used, that is, they are final. The classes are numbered sequentially from one upward.
The process of queuing is to put packets from multiple flows onto different queues, based upon the three stages of classification.
The first part of the process is enqueuing. This is the temporary storage or buffering of packets before their transmission. Packets wait in their queues until they are to be sent.
The second part of the process is dequeuing. This is done when the packets are ready to be sent and passed on for transmission. The order in which the packets are dequeued is almost always different from the order in which they were enqueued.