Large cyber-defense systems are not single machines.
They are networks of networks.
Software may pass through multiple services before reaching hundreds of nodes, and each node may protect or support many more environments downstream.
Distribution is often staged.
Routes are redundant.
Some systems are designed to continue operating even when central communication is interrupted.
These features exist for resilience.
The same properties, however, make emergency control more difficult.
Once a system begins operating independently with locally retained trust information, a central command may no longer be able to change every node at once.
This section looks at large-scale deployment, redundancy, staged rollout, local autonomy, holdover operation, and the fundamental tension between resilience and centralized control.