For software platforms, these two different sets of commands (1) and (2) exist inside a QoS policy.

• Traffic — packets — are passed into the Vyatta NOS router and reach forwarding lookup. At this stage packets are directed towards the interfaces that the traffic is to be sent out of.
• Here we do the forwarding lookup to determine what egress interface to use, then classify into flows and then schedule these flows.

For hardware platforms, these two different sets of commands (1) and (2) can both exist inside a QoS policy; or, the first set of commands (1) can exist as a separate entity while the second set of commands (2) exists inside the QoS policy.

• Everything can be inside the QoS policy, and it works much like the software router does.
• You can split the sets of commands (1) and (2) into separate entities. In this case, we split the flows on ingress, then do forwarding lookup to determine what egress interface to use, and when the packet gets to the egress interface we attach the QoS policy.

After the system determines which egress interface to use, then it can apply a QoS policy.

Each QoS policy can be applied to one or more interfaces. Each instance of a QoS policy operates independently.

Based on the classification, QoS can split the traffic into multiple streams that are directed to individual queues. Packets are then removed from these queues in an order that is determined by the queue scheduling part of the QoS policy.

Finally, the packets are sent out of the egress interface.