B does the job without taking the whole company offline. Isolating just the affected segment stops lateral movement and keeps other services up. Pretty sure that's what most incident response playbooks recommend, but let me know if you disagree.
I don't think D is right here, B fits better. Isolating the segment targets just the attack without knocking out the whole network, while D would cause major downtime. Easy trap for people thinking total shutdown is always safest. Pretty sure most IR guidelines go with B, but open to counterpoints.