Understanding NSR
Basic Concepts
- High availability (HA): supports a backup channel between the active and standby MPUs.
- Non-stop forwarding (NSF): enables uninterrupted traffic forwarding.
- Non-stop routing (NSR): enables the device to use the standby control plane to restore services immediately if an active control plane fails, whereas the control planes of neighboring nodes do not detect the fault.
- Line interface processing unit (LPU): implements forwarding plane processes.
NSR Procedure
- Batch backup: NSR is automatically enabled after the standby MPU starts. The active MPU sends routing and forwarding information in batches to the standby MPU. A batch backup is performed before a real-time backup. NSR does not perform a switchover of active and standby MPUs during the batch backup.
- Real-time backup: After the batch backup is complete, any updates to the control and forwarding planes can be backed up to the standby MPU in real time. NSR is ready to perform a switchover of active and standby MPUs to allow the standby MPU to take over traffic from the active MPU if a fault occurs.
- Switchover of active and standby MPUs: If the active MPU fails, the standby MPU detects the failure and becomes the new active MPU. The standby MPU instructs LPUs to send packets to the standby MPU itself. The switchover of active and standby MPUs is complete rapidly so that the routes between the local node and its neighboring nodes remain reachable.
NSR Switchover Process
After the NSR is enabled, the standby MPU starts and sends a message about its in-service status to the active MPU. After receiving the message, the active MPU backs up its data in batches to the standby MPU.
If the active MPU fails after the batch backup is complete, the standby MPU can become a new active MPU and restore data.
If the active MPU fails before the batch backup is complete, the standby MPU cannot become a new active MPU. The device then restarts automatically.
After the batch backup is complete, if the neighbor status or routing information changes on the active MPU, the active MPU immediately sends updated information to the standby MPU.
If software or hardware of the active MPU fails, the standby MPU detects the failure and automatically becomes the new active MPU. The new active MPU uses the backed up data to forward traffic. LPUs send the new active MPU their information, including the information that has been updated during the switchover of active and standby MPUs. Routes are reachable and traffic forwarding is uninterrupted during the switchover.
Differences Between NSF (GR) and NSR
NSF and NSR are two HA solutions. Table 5-1 describes the differences between NSF (GR) and NSR.
Difference |
NSF (GR) |
NSR |
|---|---|---|
Implementation |
The forwarding plane is not interrupted, that is, services are not interrupted. |
The control and forwarding planes are not interrupted, that is, route processing and services are not interrupted. |
Whether Other Nodes Are Dependent |
Neighboring nodes need to support NSF. NSF needs to be deployed on the entire network. |
Neighboring nodes do not need to support NSR. |
Convergence Speed |
The convergence speed is slow. When the fault is rectified, the device needs to learn the network topology and routing again. |
The convergence speed is fast. During switching, the neighbor and topology information is not lost and the neighbor relationship is not terminated. |
Impact on the System |
The load is small. |
The load is large. |