Whenever you're ready, walk me through how you'd approach this.
Walk me through your process for isolating a performance bottleneck in a distributed system.
Explain how you would establish the baseline, narrow the problem across services and dependencies, distinguish a local symptom from a systemic cause, and validate the fix. Cover the telemetry you would use, the experiments you would run, how you would manage risk during mitigation, and how you would communicate findings and next steps.