Table of Contents Link to heading

Redundant RP (Route Processor) or Supervisor Engines Link to heading

A route processor is card that performs control plane functions.
Two states are active/standby.
While failover happens, routing protocols restarts and adjacencies down. Thus. we need HA technologies to avoid downtime.

HSA (High System Availability): standby RP stays inactive until the active RP is down.
RPR (Route Processor Redundancy): standby RP is partially initialised and startup-cfg is synced.
RPR+ (Route Processor Redundancy+): standby RP is initialised and startup-cfg+running-cfg are synced. Still not synced other protocols, thus there is still downtime while failover.

SSO fully boots and initialises the standby RP.
NSF is non-configurable. It is enabled by default if SSO is enabled, hence SSO/NSF.
SSO maintains L2 forwarding by checkpointing FIB to standby RP.
However NSF doesn’t checkpoint any L3 control plane info -> adjacencies down and routes are cleared.

GR maintains L3 forwarding to a lost neighbour during an RP failover. - This length of time is called grace period. - Despite down adjacencies, packets are still forwarded.
GR is a two-way handshake. - Device failing over must be GR-capable. - Neighbouring device must at least be GR-aware, called helpers.
With GR in place, traffic continues to be forwarded even if the device is actually down (not just doing an RP failover). - To resolve this, do not set grace period too long.

NSR maintains L3 forwarding and neighbour adjacencies by checkpointing both FIB and RIB to standby RP. failover is happening.
Traffic and adjacencies operates seemlessly without awareness, but NSR really increases the workload on the router.

Info

Router(config)# redundancy Router(config-red)# mode sso