This is an overview for the outage that brought down Facebook.

A wrong command was issued and that brought down the backbone system. (No thanks to audit failure & bugs)

This caused a complete disconnection of the server connections between Facebook's data centers and the internet. No one could reach Facebook services.

Other data centers were running but they were not reachable because the Name servers (DNS) disabled advertisements. Think of those advertisements as addresses, If someone said come and visit me and s/he doesn't give you the address, you can't get to the person's location. So Facebook was not advertising it's location on the internet to anyone.

The DNS servers disabled BGP advertisements because they themselves could not speak to the backbone.

It took hours to get into the data center and the severs because of the high levels of physical and system security put in place. (No thanks to various levels of security systems, few Engineers can actually touch the rack)

When the Engineers go into the facility and access to the servers, they were able to restore the system. Always respect Engineers

When the systems were restored, they have to roll out a gradual restore process across the globe so as not to destroy the other data centers which were idle. If someone points a bright flashlight into your eyes you won't appreciate it.

Going forward Facebook has learnt some lessons.

A more efficient audit process.
How to bypass the various security systems in case of an emergency
How to minimize such failures