Goodmorning,
The network architecture in objet has been described here below:
The system is composed by 1 main site and more than 20 remote sites connected geographically by a VPLS Network provided by the National ISP. In each location a couple of Cisco Routers (3900 on Main site and 1900/2900 in other sites) have been installed. The routers of the Main site communicate with the routers of other locations via GRE tunnel that create a star topology (On the router of the Main site there are more than 20 GRE Tunnels for the connection to other remote sites). The GRE tunnel is configured as below:
interface Tunnel5
ip address 173.1.5.1 255.255.255.252
ip pim query-interval 300 msec
ip pim dense-mode
ip ospf cost 1000
ip ospf dead-interval minimal hello-multiplier 3
ip ospf retransmit-interval 1
tunnel source 172.16.0.1
tunnel destination 172.16.0.5
OSPF parameters:
router ospf 1
timers throttle spf 10 100 5000
timers throttle lsa 10 100 5000
timers lsa arrival 80
Inside VPLS Network could happen a flapping issue (E.g. a fault of 2/3 seconds each 5/6 seconds) of the geographic circuits that causes a continuous recalculation of PIM/OSPF adjacency, routes calculation, ecc... with the effect of really high CPU usage (around 100%) that block the router operativity.
My question is: in which way I could protect my network from ISP fault flappings without changing the convergence time of my network (without modifying hello packets,...)?
Awaiting your kindly reply and support
Best Regards
Claudio
