Great overview of getting started with BigQuery's speed and scale. For a follow-up, when working with large, frequently updated datasets, what's your recommended approach for structuring tables—would you lean towards partitioning by date or using clustering on other key columns for optimal performance and cost?