You need weather data for a model that uses only the last 30 days. To avoid storing unnecessary data and minimize cost, how should you design the BigQuery table?

The correct answer leverages BigQuery's native partition expiration feature, which automatically deletes partitions older than a specified duration, directly addressing the need to minimize storage and cost. Partitioning by the datetime value of the weather date ensures that entire partitions containing old weather data are removed efficiently. Incorrect options: Deleting rows with a scheduled query is less efficient and more costly than partition expiration because it involves rewriting data within existing partitions, incurring DML (Data Manipulation Language) charges and potentially affecting query performance. Partitioning by ingestion time would remove data based on when it was loaded, not when the weather event occurred, which might not align with the model's requirement for the "last 30 days" of weather data. Similar to the first incorrect option, running a scheduled query to delete rows based on a datetime column is inefficient and more expensive than using partition expiration.