Difference between NSR, NSF and GR

Latest reply: Feb 18, 2020 11:02:15 55 4 1 0

Hello guys,

As well known, both NSR, NSF and GR are used to imporve the forwarding reliability of the network devices, but do you know how to distinguish them? In this post, we are going to introduce you the difference between them. Hope you enjoy this post.

Modern high-performance routers physically separate the forwarding board and the control board in the architecture. Each board has its own processor and memory. The control board runs the routing protocol, maintains the necessary databases in the routing process, and generates a forwarding table(FIB). This forwarding table is passed to the forwarding board, which is used to guide the packets forwarding.


A great advantage of physically separating the forwarding board from the control board is that if the traffic is heavy and the forwarding board is heavily loaded, it will not affect the ability of the control board to process new routing information. Conversely, if the control board is heavily loaded (processing a large amount of new routing information), it will not affect the ability of the forwarding board to continue to forward packets at high speed.


In fact, the control board can stop working because the forwarding board can work independently after getting a copy of the forwarding table. This is called NSF (Non-Stop Forwarding): the forwarding board continues to "ignorantly" forward if the control board stops working.


Of course, this is a high-risk scenario. If the network topology changes when the control board stops working, the forwarding table of the forwarding board becomes invalid, resulting in incorrect packet forwarding. But why do we need NSF? The answer is that there are surplus control boards. NSF could prevent the packets forwarding from interrupting during switching of the main control board to the backup control board. During this switching process, the forwarding table may be invalid, and the risk during this period is acceptable. .


The shorter the switching time, the smaller the risk. Therefore, if the backup control board maintains a copy of the active control board configuration and the current system state such as the interface, the switchover will become fast because there is no need to learn in advance. This is a second-level switch. Cisco calls this process SSO (Stateful Switchover), and Juniper calls it Grace (Graceful Routing Engine Switchover).


The third level is NSR (Non-Stop Routing). So far, when the control board switches, entries losting of the adjacency table can always occur, even if SSO is used. When the active control board B stops working, each of its neighbor routers A finds that the peer-to-peer session failed. During this period, A will notify A's neighbors that the path those contain router B has failed. Of course, when the backup board is successfully started, the neighbor and the neighbor's neighbor will rebuild the path. This process can greatly interrupt the network.


The goal of NSR is to prevent, or minimize, the impact on peering sessions during the board switching.


One method to control the adjacency table during the control board switching is the Graceful Restart (GR) protocol extension. Each routing protocol has its own detailed GR extension, but they work almost the same way. When a router's control board stops working, its neighbors do not immediately report to their neighbor that the router is down, but wait for a period of time (called the grace period). If the router's control board is successfully started before this time limit, and the peer-to-peer session is rebuilt, the temporarily disabled peer-to-peer session will not affect the network outside the neighbor.


However, GR has two problems:

1. Neighbor routers must support GR protocol extensions. Switching the control board to a PE (provider edge: Provide edge device) router is very destructive, because PE is connected to many CEs (customer edge: customer edge equipment), and CE routers usually do not support GR.

2. If the router is completely damaged rather than switch, the GR will slow down the network convergence (reconvergence).

 The NSR uses an internal process to keep the backup control board aware of the routing protocol status and adjacent maintenance activities. Therefore, after the switchover is complete, the backup board can control the existing peer-to-peer session information without rebuilding a new one. Switching is transparent to the neighbors, because the NSR process is performed internally, it doesn’t require the neighbors to support any protocol extension.


That’s all for this post. If you have any problems or suggestions, please comment below, we’ll help to resolve it.


  • x
  • convention:

Created 6 days ago Helpful(0) Helpful(0)

good job
  • x
  • convention:

Created 5 days ago Helpful(0) Helpful(0)

  • x
  • convention:

Created 2 days 07:09 Helpful(0) Helpful(0)

  • x
  • convention:

Created Yesterday 11:02 Helpful(0) Helpful(0)

  • x
  • convention:


You need to log in to reply to the post Login | Register

Notice Notice: To protect the legitimate rights and interests of you, the community, and third parties, do not release content that may bring legal risks to all parties, including but are not limited to the following:
  • Politically sensitive content
  • Content concerning pornography, gambling, and drug abuse
  • Content that may disclose or infringe upon others ' commercial secrets, intellectual properties, including trade marks, copyrights, and patents, and personal privacy
Do not share your account and password with others. All operations performed using your account will be regarded as your own actions and all consequences arising therefrom will be borne by you. For details, see " Privacy."
If the attachment button is not available, update the Adobe Flash Player to the latest version!
Login and enjoy all the member benefits

Login and enjoy all the member benefits