Distinguish vertical scaling (choosing a larger worker size for memory-bound, large-payload, or CPU-intensive processing) from horizontal scaling (adding more workers for higher concurrency and availability), and select the correct scaling axis for a stated bottleneck.