>hmm why did they have to change the server? Why was it so global?
this sounds to me like literal truth that is completely misleading.
how does a single configuration change on a server bring about total loss of service?
one way is for that change to be the start of a process that propogates changes across the whole (or some significant subset) of the environment. and this is exactly how DC infrastructure environments like FBs are managed in terms of feature/functionality and software/device config.
some major change to functionality, or some massive data wipe. something that creates a huge backlog of high priority tasks.
who the fuck knows.