My question is the same as the title and I wrote down what I understood below. If there is something wrong, could you explain the wrong part and contents?
1. sso has active RP and standby RP
2. If RP failed, active RP dies and standby RP takes its place.
3. Because sso forwards Layer 2 communications, Layer 2 protocols are routed
4. However, sso does not forward Layer 3 traffic, so it cannot be routed
5. NSF copies the FIB of the active RP and informs the standby RP.
6. NSF is a friend who handles the three-tier layer, but this friend only runs step 5 with the local process and does not do anything else.
7. So GR or NSR is required GR should also be set on the other router GR sets a grace period to prevent neighbors from deleting peers during the grace period Communication is possible because the peer is not deleted and GR forwards it Active RP is restored and normal communication is restored within the grace period
8. NSR plays a similar role to GR, but I personally think it's a much better way → one of feature "adjacency", which means that neighbors do not delete adjacencies, so they may not know if RP failed in that router during which time active RP is recovered
