Cisco Nexus 9000 Series Configuration Manual page 23

Overview
Policing
Policing is the monitoring of data rates for a particular class of traffic. The device can also monitor associated
burst sizes.
Single-rate policers monitor the specified committed information rate (CIR) of traffic. Dual-rate policers
monitor both CIR and peak information rate (PIR) of traffic.
Queuing and Scheduling
The queuing and scheduling process allows you to control the bandwidth allocated to traffic classes so that
you achieve the desired trade-off between throughput and latency.
You can apply weighted random early detection (WRED) to a class of traffic, which allows packets to be
dropped based on the QoS group. The WRED algorithm allows you to perform proactive queue management
to avoid traffic congestion.
You can shape traffic by imposing a maximum data rate on a class of traffic so that excess packets are retained
in a queue to smooth (constrain) the output rate. In addition, minimum bandwidth shaping can be configured
to provide a minimum guaranteed bandwidth for a class of traffic.
You can limit the size of the queues for a particular class of traffic by applying either static or dynamic limits.
ECN can be enabled along with WRED on a particular class of traffic to mark the congestion state instead of
dropping the packets.
Sequencing of QoS Actions
The following are the three types of policies:
• network qos—Defines the characteristics of QoS properties network wide.
• qos—Defines MQC objects that you can use for marking and policing.
• queuing—Defines MQC objects that you can use for queuing and scheduling.
Note The default type of policy is qos.
The system performs actions for QoS policies only if you define them under the type qos service policies.
Sequencing of Ingress Traffic Actions
The sequence of QoS actions on ingress traffic is as follows:
1. Classification
2. Marking
3. Policing
Cisco Nexus 9000 Series NX-OS Quality of Service Configuration Guide, Release 7.x
Policing
9