Which combination of practices most effectively reduces watsonx.data operational costs?

Cost is minimized by: pausing idle engines (stops compute charges); right-sizing engines per workload (no over-provisioning); and compacting data files (fewer object storage requests during scans). The incorrect options ("Running all engines at maximum size continuously to avoid startup latency", "Storing all data in CSV format to avoid Parquet conversion costs", "Using a single large Presto engine for all workload types to reduce complexity") are distractors that don't fully capture the concept described.