What Is Intelligent Traffic Policing?
Intelligent traffic policing (ITP) is a technology that dynamically adjusts the rate limit. It is implemented based on hierarchical CAR HQoS. By dynamically monitoring interface bandwidth and packet loss, the device configured with intelligent traffic policing automatically adjusts the rate limiting policies for high-priority and low-priority services, implementing dynamic on-demand allocation of bandwidth resources.
Why Is Intelligent Traffic Policing Required?
Traditional traffic policing limits the rate based on the bandwidth. However, during peak hours, the available bandwidth fluctuates within a certain range, which cannot guarantee the experience of high-priority services. As shown in the following figure, during peak hours, the available bandwidth of an external device deteriorates to 90 Mbit/s, which cannot guarantee the bandwidth required by services. In this case, the enterprise gateway limits the rate based on the bandwidth, that is, 70 Mbit/s for high-priority services and 30 Mbit/s for low-priority services. As the external device is the link bottleneck (90 Mbit/s < 100 Mbit/s), the rate limit on the enterprise gateway does not take effect. As a result, packet loss occurs on the external device for high-priority services (the rate decreases from 70 Mbit/s to 65 Mbit/s), affecting user experience of high-priority applications.
The available bandwidth fluctuates. If the rate limit is too low, the bandwidth utilization is insufficient. If the rate limit is too high, local QoS does not take effect. To address this issue, we innovatively propose intelligent traffic policing, which can intelligently guarantee the bandwidth of high-priority service traffic and allocate the remaining bandwidth to low-priority service traffic.
Bandwidth deterioration
Intelligent traffic policing is a technology that dynamically adjusts the rate limit. It moves the congestion bottleneck to an enterprise gateway and uses the WAN available bandwidth evaluation technology and dynamic rate limit adjustment algorithm to automatically optimize the rate limiting policy on the enterprise gateway. In this way, backpressure can be performed on low-priority services, and high-priority services are preferentially guaranteed.
Currently, there is a large amount of TCP application traffic on the live network. In addition, TCP dynamically adjusts the application bandwidth based on the sliding window mechanism. Therefore, for low-priority TCP applications, the gateway identifies the applications and limits the rate. The bandwidth of low-priority TCP applications in the inbound direction of the device is reduced, so that the bandwidth can be reserved for high-priority applications. In this way, dynamic and autonomous traffic management and control can be implemented, manual intervention can be reduced, bandwidth resource utilization can be maximized, and high-priority applications can be guaranteed.
Currently, intelligent traffic policing is supported only in SD-WAN scenarios and requires a license.
What Are the Application Scenarios of Intelligent Traffic Policing?
Currently, intelligent traffic policing is supported only in SD-WAN scenarios. As shown in the figure, after hierarchical CAR HQoS is applied to the inbound interface of CPE1, multi-level traffic policing can be implemented in the inbound direction of CPE1. However, during peak hours, the actual available bandwidth fluctuates within a certain range. As a result, rate limiting in the inbound direction of CPE1 becomes invalid, affecting the experience of high-priority applications on CPE2 and CPE3. Therefore, after hierarchical CAR HQoS is applied, you can configure intelligent traffic policing. When the incoming traffic rate of CPE1 reaches the configured bandwidth threshold, intelligent traffic policing is enabled to preferentially guarantee the bandwidth of high-priority service traffic and allocate the remaining bandwidth to low-priority service traffic.
- Create an intelligent traffic policing policy on CPE1.
- Apply the intelligent traffic policing policy to the traffic behavior of the parent traffic policy on CPE1.
Network diagram of the hierarchical CAR scenario model
How Does Intelligent Traffic Policing Work?
After intelligent traffic policing is enabled on the device, the system evaluates the available bandwidth and dynamically adjusts the available bandwidth of an interface to preferentially guarantee the bandwidth of high-priority service traffic. The following figure shows the working process of intelligent traffic policing.
- Configuring hierarchical CAR HQoS: You can configure application classification and rate limiting through hierarchical CAR HQoS and apply an intelligent traffic policing policy, specify high-priority and low-priority application queues in hierarchical CAR, and configure the bandwidth threshold and interval for intelligent traffic policing.
- Traffic identification and classification: Incoming traffic enters different queues based on the priority after application identification.
- Bandwidth evaluation: The device periodically collects traffic statistics on the inbound interface, determines whether packet loss occurs on the interface, and evaluates the available bandwidth.
- Dynamic adjustment: If packet loss occurs on the inbound interface, the link bottleneck is on the local device. In this case, the device increases the rate limit and detects more available bandwidth. If no packet loss occurs on the inbound interface, the link bottleneck is not on the local device. In this case, the local device limits the bandwidth of low-priority services through the backpressure mechanism and periodically detects the available bandwidth.
- Function disabled: When the incoming traffic rate on the inbound interface is lower than the threshold and no packet loss occurs for low-priority applications, intelligent traffic policing is disabled.
Working process of intelligent traffic policing
