Diagnose I/O, Memory, and CPU Constraints

# Diagnose I/O, Memory, and CPU Constraints Describe situations in which I/O, memory, and CPU were each the limiting resource. For each case, explain how you measured the bottleneck, which change you made, and how you established that performance improved without breaking correctness. ### Constraints & Assumptions - Keep employers, systems, and customers anonymous where necessary. - Do not infer a bottleneck from utilization alone; relate measurements to queueing, latency, throughput, and workload. - Include at least one rejected hypothesis or trade-off. ### Clarifying Questions to Ask - Is the interviewer asking for three separate examples or one system that moved between constraints? - Which service-level or batch-completion target mattered? ```hint Describe the evidence chain The strongest story connects a symptom to measurements, a falsifiable hypothesis, a bounded change, and a before/after result. ``` ### What a Strong Answer Covers - Correct signals for I/O wait, memory pressure, and CPU saturation. - A concrete evidence chain, intervention, and measured outcome for each of the three limiting resources. - Workload-aware profiling rather than premature optimization. - A targeted mitigation with costs and capacity limits. - Validation, regression safety, and monitoring after the change. ### Follow-up Questions - How can adding concurrency make each of these constraints worse? - What would you measure before deciding to scale horizontally?