The noisy neighbor problem happens when one workload on shared infrastructure consumes disproportionate CPU, memory, storage, or network capacity and slows down other tenants or services. In cloud systems, this is a common source of unpredictable latency because applications may share the same host, cluster, storage backend, or network path even when their code has not changed.
In cyber security, the issue matters because performance degradation can look like a normal outage, hide active abuse, or be used as a denial-of-service-style pressure tactic in multi-tenant environments. Defenders watch for sudden jumps in tail latency, queue depth, and resource saturation, then correlate them with traces, logs, and node-level metrics. Mitigations include strict resource limits, workload isolation, autoscaling, placement controls, dedicated instances, and continuous telemetry to detect contention before users feel it.