для Cisco Cisco IOS Software Release 12.0(23)S

Usually, when a networking device restarts, all routing peers of that device detect that the device went 
down and then came back up. This transition results in what is called a routing flap, which could spread 
across multiple routing domains. Routing flaps caused by routing restarts create routing instabilities, 
which are detrimental to the overall network performance. Cisco NSF helps to suppress routing flaps in 
SSO-enabled devices, thus reducing network instability.

Cisco NSF allows for the forwarding of data packets to continue along known routes while the routing 
protocol information is being restored following a switchover. With Cisco NSF, peer networking devices 
do not experience routing flaps. Data traffic is forwarded through intelligent line cards or dual 
forwarding processors (FPs) while the standby RP assumes control from the failed active RP during a 
switchover. The ability of line cards and FPs to remain up through a switchover and to be kept current 
with the Forwarding Information Base (FIB) on the active RP is key to Cisco NSF operation.

Cisco NSF always runs together with SSO. This section provides some background information on the 
SSO feature. 

In specific Cisco networking devices that support dual RPs, SSO establishes one of the RPs as the active 
processor while the other RP is designated as the standby processor, and then synchronizes information 
between them. A switchover from the active to the standby processor occurs when the active RP fails, is 
removed from the networking device, or is manually taken down for maintenance. 

In networking devices running SSO, both RPs must be running the same configuration so that the 
standby RP is always ready to assume control following a fault on the active RP. The configuration 
information is synchronized from the active RP to the standby RP at startup and whenever changes to 
the active RP configuration occur. Following an initial synchronization between the two processors, SSO 
maintains RP state information between them, including forwarding information.

During switchover, system control and routing protocol execution is transferred from the active 
processor to the standby processor. The time required by the device to switch over from the active to the 
standby processor ranges from just a few seconds to approximately 30 seconds, depending on the 
platform. 

SSO supported protocols and applications must be high-availability (HA)-aware. A feature or protocol 
is HA aware if it maintains, either partially or completely, undisturbed operation through an RP 
switchover. For some HA aware protocols and applications, state information is synchronized from the 
active to the standby processor. For Cisco NSF, enhancements to the routing protocols (Cisco Express 
Forwarding, or CEF; Open Shortest Path First, or OSPF; Border Gateway Protocol, or BGP; and 
Intermediate System-to-Intermediate System, or IS-IS) have been made to support the HA features in 
SSO.

For more information on SSO, see the section “Related Documents.” 

Cisco NSF is supported by the BGP, OSPF, and IS-IS protocols for routing and by Cisco Express 
Forwarding (CEF) for forwarding. Of the routing protocols, BGP, OSPF, and IS-IS have been enhanced 
with NSF-capability and awareness, which means that routers running these protocols can detect a 
switchover and take the necessary actions to continue forwarding network traffic and to recover route 
information from the peer devices. The IS-IS protocol can be configured to use state information that 
has been synchronized between the active and the standby RP to recover route information following a 
switchover instead of information received from peer devices.