Classification and marking is an essential component to establishing end-to-end quality of service (QoS), which...
ensures a reliable voice over IP (VoIP) infrastructure. Classification and marking is a system of identifying packets or traffic flows and assigning certain parameters within the packet headers in order to group them. Once the traffic is "identified," it can be marked or "colored" into groups so that QoS policies can be applied to them.
Packets can be identified by their MAC address, VLAN IDs, source or destination IP addresses, their class of service (CoS), their ingress or egress interfaces, their inclusion or exclusion within an access list, or their associated application or protocol. For example, HTTP, SNMP, and FTP traffic can be marked in order to assign quality of service parameters that could govern the traffic's available bandwidth or queue delay.
An end-to-end QoS model requires the ability to mark a traffic stream into one of a group of classifications. Each classification incorporates a QoS service level. The IETF created a model using the type of service (ToS) field in the IP header to serve this purpose. We'll discuss two QoS models: Integrated and Differentiated Services.
The Integrated Services QoS Model is one that permits the end-hosts to transmit their required QoS service levels to the network. This model uses the IP Precedence to mark traffic into eight possible groups. The IP Precedence bits are the first three bits of the ToS Field.
The lower numbered precedence packets (000) have lower priority to the higher numbered precedence packets (111) and should be discarded during times of network congestion. Once packets are marked using IP Precedence, the QoS network infrastructure will understand their priority and apply the appropriate QoS parameters to the flow. For example, real time traffic (voice) might be assigned an IP Precedence value of 111. The required high availability, low delay QoS parameters could then be assigned each flow represented with that IP Precedence value.
The Differentiated Services QoS Model adopts the requirement for simple and coarse methods of grouping traffic into different classes, and applying QoS service levels to those classes. Similar to Tos/IP Precedence, packets are first divided into classes by marking the type of service (ToS) byte in the IP header. However, the Differentiated Services Code Point (DSCP) is the first six bits within the IPv4 ToS Octet. With DSCPs 0 through 63 available, a maximum of 64 different aggregates/classes can be supported.
The foundation of the DiffServ architecture involves the classification and conditioning (as required) of traffic as it enters the network and/or as it leaves the network. Within the network infrastructure, packets are forwarded based on the per-hop behavior related to the DS Codepoint. Because of the intense efforts required to determine the appropriate class of traffic for packets, it is ideal to minimize the number of classification occurrences within the network infrastructure. By marking the traffic at the network edge, the remaining network devices along the forwarding path are allowed to quickly determine the proper grouping of a given traffic flow.
Richard Parsons (CCIE#5719) is a Principal Architect for AT&T with a focus on network planning, design, and implementation. He has built a solid foundation in networking concepts, advanced troubleshooting, and monitoring in areas such as optical, ATM, VoIP, routed, routing, and storage infrastructures. Rich resides in Atlanta GA, and is a graduate of Clemson University. His background includes consulting positions at International Network Services, Lucent, and Callisma.