This chapter contains a general overview of Quality of Service (QoS) functionality and terminology.

QoS can be used to give certain traffic priority over other traffic. Without QoS, all traffic in a network has the same priority and chance of being delivered on time. If congestion occurs, all traffic has the same chance of being dropped. With QoS, specific network traffic can be prioritized to receive preferential treatment. In turn, a network performs more predictably, and utilizes bandwidth more effectively.

QoS is based on DiffServ architecture, which stipulates that individual packets be classified upon entry into a network. Classification information can be carried in the Layer-3 IP packet header or the Layer-2 frame. IP packet headers carry the information using 6-bits from the deprecated IP type of service (TOS) field. Layer-2 802.1Q frames carry the information using a 2-byte Tag Control Information field. All switches and routers accessing the Internet depend on class information to give the same forwarding treatment to packets with the same class information, and give different treatment to packets with different class information. A packet can be assigned class information, as follows:

Class information can be used by switches and routers along a path to limit the amount of allotted resources per traffic class. Per-hop behavior is an individual device’s behaviour when handling traffic in the DiffServ architecture. An end-to-end QoS solution can be created if all devices along a path have consistent per-hop behavior.

Quality of Service (QoS) provides preferential treatment to specific traffic, possibly at the expense of other traffic. Without QoS, Qumran offers best-effort service to each packet, however, this may cause unpredictable network behavior. Implementing QoS in a network makes performance more predictable and bandwidth utilization more effective.

QoS design in Qumran complies with IETF-DiffServ and IEEE 802.1p standards. A typical QoS model deployment is based on the following elements:

Following is a brief description of terms and concepts used to describe QoS.

Access control lists (ACLs) classify traffic with the same characteristics.

Class of Service (CoS) is a 3-bit value used to classify the priority of Layer-2 frames upon entry into a network. QoS classifies frames by assigning priority-indexed CoS values to them, and gives preference to higher-priority traffic.CoS values range from zero to seven, seven being the highest priority.

DSCP Value Differentiated Services Code Point (DSCP) is a 6-bit value used to classify the priority of Layer-3 packets upon entry into a network. DSCP values range from 0 to 63, 63 being the highest priority, 0 being best-effort traffic.

Classification distinguishes one kind of traffic from another by examining the fields in the packet. Classification is enabled only if QoS is globally enabled on the switch. By default, QoS is globally disabled, thus, no classification occurs. Classification occurs on an ingress physical port. Classification can be based on QoS ACLs, or class maps and policy maps.

Policing can occur on ingress interfaces. Policer limits the bandwidth consumed by a traffic flow with the results given to the marker. The two types of policers:

Marking determines how to handle a packet when it is out of profile. It assesses the policer and the configuration data to determine the action required for the packet, and then handles the packet using one of the following methods:

Marking can occur on ingress and egress interfaces.

Queuing maps packets to a queue. Each egress port can accommodate up 8 queues, prioritized as 0 lowest and 7 highest. The tagged packet incoming priority can be mapped to one of the 8 queues obtained from the filtering mechanism result. The untagged packet priority is also obtained from the filtering mechanism result. After the packets are mapped to a queue, they are scheduled.

Scheduling forwards or conditions packets using one of the following methods:

A class map names and isolates specific traffic from other traffic. The class map defines the criteria used to match against a specific traffic flow to classify it further. The criteria can include:

If there is more than one type of traffic to be classified, another class map can be created under a different name. After a packet is matched against the class-map criteria, it is further classified using a policy map.

A policy map specifies on which traffic class to act. This can be implemented as follows:

Policy maps have the following attributes:

This section explains the implementation and configuration details of QoS:

Data — Packets can be network-to-network traffic or traffic from CPU. Network-to-network packets are considered data traffic and QoS parameters can be applied on data traffic.

Control — Packets to and from the CPU are considered as control traffic. Incoming control traffic is prioritized based on ingress map and are sent to their own designated CPU queues. Each CPU queue has a fixed rate limit to guard the CPU. Outgoing control traffic will always be sent on highest priority queue (Q7) on the data port.

The device supports modifying the packets header IEEE 802.1p user priority or IP-DSCP.

Every data packet is assigned a set of QoS attributes that can be modified in several stages of the ingress pipeline engine.

The ingress pipeline engine also contains a QoS Remark option for L3 traffic that can modify the initial QoS attributes of the packet.

Color and Drop Precedence relation: Green has the lowest drop precedence, Yellow has a higher drop precedence, and Red has the highest drop precedence. See Table 1-1.