Brocade Multi-Service IronWare Routing Configuration Guide (Supporting R05.6.00) - Manual (Page 445)

Multi-Service IronWare Routing Configuration Guide 417
53-1003033-02
IPv6 Non stop routing and graceful restart
IPv6 Non stop routing and graceful restart
At times, routers may need to restart or may undergo failover. Traditionally during a restart or
failover, sessions with the restarting devices are tore down and re-established. Traffic is disrupted
due to route deletion and addition in the forwarding plane. Graceful Restart (GR) and Non Stop
Routing (NSR) are two different mechanisms to prevent routing protocol re-convergence during a
processor switchover.
When Graceful Restart is used, peer networking devices are informed, via protocol extensions that
the router is undergoing a restart condition. Peer devices, known as “helper” devices, will continue
to forward to the restarting router until a “grace period”, within which the adjacency is
re-established.
When Non Stop Routing is used, peer networking devices have no knowledge of any event on the
router that is switching over. All information needed to continue the routing protocol peering state is
transferred to the standby processor so it can continue immediately upon a switchover. Since NSR
does not require the help of neighboring routers during restart, NSR capable routers can be
deployed independently in an existing network.
Limitations
• Configuration events that occur at the same time as the switchover may get lost are lost due to
the CLI synchronization.
• Neighbor, interface, or NSSA translation state changes 'close' to and during the switchover will
not be handled.
• Due to the core-reset of the LP, dead-timers below 40 seconds are not supported.
• Number of neighbors supported may be limited depending on how many packets LP can
send upon completion of the core-reset, due to competition with LP-sync-updates to get
OSPF neighbor packets sent out.
• Traffic counters will not be synced. Neighbor and LSA DB counters will be recalculated on
Standby during sync.
• There may be a slowdown of LSA acking due to the wait for the ack from Standby before acking
the received LSAs.
• OSPF Database Overflow condition for External LSAs - depending on the sequence of
redistribution or new LSAs (from neighbors), the LSAs accepted within the limits of the
database may change upon switchover.
• The NSR hitless failover event may not be completely transparent to the network as after
switchover additional flooding related protocol traffic will be generated to the directly
connected neighbors.
• OSPF Startup Timers will not be applied upon NSR switchover.