Comment by pm2222
5 years ago
internal tools to bring back the backbone relied on DNS which relied on the backbone being up
So are you referring to same DNS servers sitting outside the backbone at the various PoPs? I'd imagine some internal DNS servers which stays in the backbone at use here, unless of course the FB engineers themselves were disconnected from those internal DNS servers.
I don't recall how internal DNS was setup (and determining from the outside isn't really possible), but there were comments in the incident report that DNS being unavailable made it harder to recover.