Hey!
You always (should) have an APIC Cluster of at least 3 APIC Nodes..
If one APIC dies, one of the other takes over. That should be transparent for you.
Only issue I ever ran into:
For every VMM integration, one particular (HEALTHY!) APIC Node is responsible.
One time I had the issue that one VMM integration out of a few stopped working.. after creating new EPGs and assinging the VMM Domain, no new port-groups showed up in vCenter.
The issue was that somebody screwed up the firewall rules and blocked the access of one particular APIC Node to the vCenter
Hope that hepls.
BR
Jules
