Abstract - Concepts from the domain of fault-tolerant computing cannot be merely adopted for cyber defense; instead they have to be adapted. Defense of cyberspace is challenging. The seemingly endless breadth of cyberspace coupled with the technological depth of its composition can divide defensive approaches to be either overarching or highly specific. In order to abstract away details for the purpose of tractability, overarching approaches can suffer because simplistic models for threats, vulnerabilities, and exploits tend to yield defenses that are too optimistic. Approaches that deal with specific threats, vulnerabilities and exploits may be more credible but can quickly lose their meaningfulness as technology changes. Whether approaches are near-or-far term, we see that two underlying attributes remain essential: the ability to survive and the ability to fight-through. When a cyber defense’s ability to predict, prevent, avoid, and detect an attack are outmaneuvered and information systems face impending loss of critical services, a fight-through capability must remain; otherwise restoration of those services may come too late to emerge undefeated. The task of "protecting the protector" drives us to create a fight-through capability that is hardened and heavily defended in cyberspace; however, these attributes alone become an instantiation of a "Maginot Line". Such a strict bastion mentality should be replaced by one that advocates agility. Our goal then becomes more realistic: to design a fightthrough capability that can absorb punishment and reacts by rebounding to serve as the basis for restoration of critical services. We liken the fight-through problem to an Observe, Orient, Decide, and Act (OODA) loop. Redundancy, as the underpinning of fault tolerance, is strategically placed to counter an attacker’s optimal strategies. The aim of a fight-through OODA loop is to outperform the adversary’s OODA loop.