Whenever you're ready, walk me through how you'd approach this.
How would you debug a performance bottleneck in a distributed data pipeline?
Explain how you would isolate the slow stage, distinguish infrastructure limits from data-related issues, and validate the fix. Cover the metrics, logs, query profiles, and tracing information you would inspect across ingestion, processing, data warehouse loading, and orchestration. Discuss how you would handle common causes such as data skew, insufficient parallelism, inefficient transformations, backpressure, warehouse queuing, and storage layout problems.