Cisco WS-SUP32-GE-3B Software Configuration Guide - Page 84

Understanding NSF with SSO Supervisor Engine Redundancy Chapter 5 Configuring NSF with SSO Supervisor Engine Redundancy Note If you configure IETF on the networking device, but neighbor routers are not IETF-compatible, NSF will abort following a switchover. If the neighbor routers on a network segment are not NSF-aware, you must use the Cisco configuration option. The Cisco IS-IS configuration transfers both protocol adjacency and link-state information from the active to the redundant supervisor engine. An advantage of Cisco configuration is that it does not rely on NSF-aware neighbors. IETF IS-IS Configuration As quickly as possible after a supervisor engine switchover, the NSF-capable router sends IS-IS NSF restart requests to neighboring NSF-aware devices using the IETF IS-IS configuration. Neighbor networking devices recognize this restart request as an indicator that the neighbor relationship with this router should not be reset, but that they should initiate database resynchronization with the restarting router. As the restarting router receives restart request responses from routers on the network, it can begin to rebuild its neighbor list. After this exchange is complete, the NSF-capable device uses the link-state information to remove stale routes, update the RIB, and update the FIB with the new forwarding information; IS-IS is then fully converged. The switchover from one supervisor engine to the other happens within seconds. IS-IS reestablishes its routing table and resynchronizes with the network within a few additional seconds. At this point, IS-IS waits for a specified interval before it will attempt a second NSF restart. During this time, the new redundant supervisor engine will boot up and synchronize its configuration with the active supervisor engine. The IS-IS NSF operation waits for a specified interval to ensure that connections are stable before attempting another restart of IS-IS NSF. This functionality prevents IS-IS from attempting back-to-back NSF restarts with stale information. Cisco IS-IS Configuration Using the Cisco configuration option, full adjacency and LSP information is saved, or checkpointed, to the redundant supervisor engine. Following a switchover, the newly active supervisor engine maintains its adjacencies using the check-pointed data, and can quickly rebuild its routing tables. Note Following a switchover, Cisco IS-IS NSF has complete neighbor adjacency and LSP information; however, it must wait for all interfaces to come on line that had adjacencies prior to the switchover. If an interface does not come on line within the allocated interface wait time, the routes learned from these neighbor devices are not considered in routing table recalculation. IS-IS NSF provides a command to extend the wait time for interfaces that, for whatever reason, do not come on line in a timely fashion. The switchover from one supervisor engine to the other happens within seconds. IS-IS reestablishes its routing table and resynchronizes with the network within a few additional seconds. At this point, IS-IS waits for a specified interval before it will attempt a second NSF restart. During this time, the new redundant supervisor engine will boot up and synchronize its configuration with the active supervisor engine. After this synchronization is completed, IS-IS adjacency and LSP data is check-pointed to the redundant supervisor engine; however, a new NSF restart will not be attempted by IS-IS until the interval time expires. This functionality prevents IS-IS from attempting back-to-back NSF restarts. Catalyst Supervisor Engine 32 PISA Cisco IOS Software Configuration Guide, Release 12.2ZY 5-6 OL-11439-03