> I am building redundancy within that setup. I mean it will be software > based BGP so if hardware if fried up, it will break BGP session and pull > off routes anyway and for cases like DNS server (software) failure, I will > monitor it via simple bash script which can turn bgp daemon down. So once > it is off, routing tables should take it to different node.
i have a tee shirt which says All Devices Break Two Devices Will Break at Once Global BGP Never Converges randy, a co-author of 2182