Computing Library › Security & Safety-Critical Computing
Security & Safety-Critical Computing

Defense in Depth

Layered controls ensure that the failure of any single defense does not lead to compromise, because an attacker must defeat several in sequence.

The Core Idea

Defense in depth assumes that every individual control can fail: a firewall can be misconfigured, a password can be phished, a patch can be missed. Rather than trusting one strong barrier, it stacks independent barriers so that breaching one still leaves the attacker facing others. The concept is borrowed from military fortification and physical safety engineering, where multiple redundant containment layers protect against a single point of failure.

Layers in Practice

Kronos motion — safety factor

Independence Is What Matters

Layers only help if they fail independently. Two firewalls from the same vendor with the same flaw are one layer wearing two coats. Effective depth mixes mechanisms so that the weakness defeating one control does not defeat the next. A phished credential should still meet multi-factor authentication; a bypassed network control should still meet host hardening.

Diversity and Redundancy

Depth pairs naturally with diversity: different technologies, different vendors, different failure modes. In safety engineering this is called diverse redundancy, and it guards against common-cause failures that would take out identical components at once.

Fusion Context

For a fusion plant, defense in depth spans both cyber and physical. A command to a magnet supply must pass network segmentation, authenticated control paths, and finally an independent machine-protection interlock that acts on physics limits regardless of what the supervisory software believes. The safety case for the Hyperion breeder and burner designs treats the last protective layer as non-bypassable by design, so that no cyber compromise alone can drive the machine past a hardware limit.