Fault Containment Boundary

Fault Containment Boundary is the architectural principle that failures in one component of a system should not propagate to other components, limiting the \'\'blast radius\'\' of any single fault. It is the systems-level equivalent of a firebreak: not the prevention of fire, but the prevention of \'\'conflagration\'\'. The boundary can be physical (separate power supplies), temporal (circuit breakers that trip before overload spreads), organizational (independent teams with separate budgets), or computational (\'\'process-per-connection\'\' in databases, where each client runs in an isolated process).

The principle appears across scales. In electrical engineering, fuses and circuit breakers are fault containment boundaries. In distributed systems, bulkheads and circuit breaker patterns prevent cascading failures. In organizations, \'\'Viable System Model\'\' System 2 (coordination) exists partly to contain oscillations in System 1 (operations) before they destabilize the whole. The 2008 Financial Crisis can be read as a failure of fault containment: the boundaries between investment banks, insurers, and rating agencies were porous, and risk propagated through channels that no single regulator could see.

The design tension is between containment and efficiency. Tight boundaries isolate faults but also prevent \'\'resource sharing\'\' and coordination. Loose boundaries enable efficiency but create paths for failure propagation. The optimal boundary is not a constant; it depends on the \'\'failure mode distribution\'\' of the system. Systems with frequent, small faults need tight boundaries. Systems with rare, catastrophic faults need boundaries designed for graceful degradation rather than isolation.