Configuring NQA for PBR
As network service requirements vary widely and service data is stored in a centralized manner, network services increasingly depend on high link quality. PBR allows network administrators to make user-defined policies to change packet forwarding paths based on source and destination addresses of packets. PBR does not have a dedicated detection mechanism. If a non-directly connected link becomes faulty, the redirection next-hop address in PBR will not change automatically and needs to be changed by network administrators. This process delays link switchovers and may cause lengthy service interruptions.
To address this issue, an efficient method is required to detect the redirection link quality in PBR. Smart policy-based routing (SPBR) is such a method that is implemented using NQA for PBR. NQA for PBR uses an NQA test instance to detect the link status and determine whether PBR takes effect according to the test results. SPBR actively detects the link quality and matches service requirements to select an optimal link to forward service data, preventing network blackholes and flappings.
- If no NQA test instance exists or the NQA test instance type is not ICMP, link detection and the PBR configuration will fail.
- If an NQA test instance is bound to PBR successfully, and link detection is normal, PBR takes effect.
- If an NQA test instance is bound to PBR successfully, but link detection fails continuously, PBR automatically becomes ineffective.
- If an NQA test instance is bound to PBR successfully, and the link fault is rectified, PBR automatically takes effect.
For details about NQA, see "NQA Configuration - Principles" in the Configuration Guide - Network Management and Monitoring Configuration.
Only an ICMP NQA test instance can be bound to PBR.
The NQA test instance bound to PBR cannot be deleted and its type cannot be changed.
Application Scenario
As shown in Figure 11-1, each access switch is connected to N clients. Network administrators configure PBR on SwitchA to redirect packets from the previous forwarding path SwitchA→SwitchB→RouterB (RouterA) to a new path SwitchA→SwitchC→RouterC (RouterA) and bind an NQA test instance to PBR to detect the link status of the new path. If the NQA test instance finds that the link of the new path is working properly, packets can be forwarded normally. If the NQA test instance finds that the link of the new path is Down and the number of detection times exceeds the specified value, the PBR configuration automatically becomes ineffective, and packets are forwarded along the previous forwarding path.