Fault isolation is the specific fault tolerance requirement concerned with containing the impact of a failure to the smallest possible scope, using architectural boundaries, such as separate fault domains, network segmentation, or modular component design, so that a failure in one part of the system cannot cascade or propagate to affect unrelated components elsewhere in the environment. This precisely matches the stated goal of limiting failure impact and preventing it from spreading, which is the defining purpose of fault isolation as a design principle, distinct from other fault tolerance requirements that address different aspects of resiliency. Load balancing (A) distributes workload across multiple resources to optimize utilization and avoid overloading any single component, a capacity-management technique rather than a containment mechanism for failures that have already occurred. Redundancy implementation (B) provides backup or duplicate components so that a workload can continue running if one component fails, addressing continuity of service rather than specifically preventing the failure's effects from spreading to unrelated parts of the system. Fault recovery (D) concerns restoring a failed component or system back to normal operation after the failure has occurred, a remediation activity distinct from the containment goal described. Eliminating single points of failure (E) is a broader architectural goal achieved partly through redundancy, related to but not synonymous with the specific isolation mechanism the question describes. Fault isolation is correct.

Reference topic: Fault Tolerance Techniques - Fault Isolation as a Core Requirement.