Applying the QoS Model

Applying the QoS Model
The first stage in determining how the switches and routers in the network will prioritize traffic
is the classification process. Essentially, the idea is to somehow mark traffic with an indication
that it should be treated differently from packets with dissimilar marking.
The second stage is traffic policing. This is the process whereby a switch/router determines
whether the frame/packet matches the preconfigured profiles. Packets that exceed specified limits
are considered to be nonconforming. The policing process specifies the action to take for
packets by either setting bandwidth limits for conforming traffic, or by dropping or remarking
nonconforming traffic.
The third stage is to actually mark the frame/packet. Data can be marked at layer 2 (in the
802.1p header) or at layer 3 (inside the IP header), depending upon the device. Switches that
operate at layer 3 are able to mark at either layer, but switches operating purely at layer 2 are
able to mark only the frame.
So if the switch is a layer 3 switch, we have the option of forwarding a packet with QoS.
Then, using the general principles of traffic types, we need to “map” the traffic type to a TOS
or DiffServe number. After the packet has been through the classification, policing, and marking
processes, it is assigned to the appropriate queue before exiting the switch. If the switch has
received the packet inside an 802.3 frame with 802.1p priority specified, this process may be
automated. If not, then we must map it manually.
Finally, the packet must be forwarded out of a shared output buffer onto the media toward
the next hop. This is usually accomplished by establishing a queuing process and placing traffic
into different queues according the policies defined earlier.