If the active card fails,
the backup is thus able to assume the active role without its neighbours being
aware of the switchover. One disadvantage of this approach is that state synchronisation
requires significant additional CPU resources to operate.
?– Nonstop forwarding An alternative to ensuring continued operation of routing
and signalling protocols during a control card switchover is to enhance the protocols
so that when a switchover occurs, the failed node is given time to recover by its
neighbours and is able to ask those neighbours to refresh its state. Graceful Restart
extensions have been implemented for OSPF, IS-IS, RSVP, LDP, and BGP to facilitate
this. One disadvantage of this approach is that for a router to perform a graceful
restart its neighbours must also implement the protocol extensions.
Another area of concern for operators is the need to upgrade node software periodically
to take advantage of new features or to fix bugs. If this occurs in the core of the
network, it is possible to steer LSPs away from the node being upgraded for the duration
of the upgrade. However, as with node switchovers, if an edge node needs to be
rebooted to activate a software upgrade, this will impact customers directly connected
to that node.
Pages:
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972