問題文
A directory of Parquet files totals 30 GB on disk. The team must decide whether the data fits on a 48 GB device. Which procedure gives a usable answer?
選択肢
- Load everything and see whether it fails.
- Multiply the 30 GB by the compression ratio reported by the writer, which gives the exact in-memory size.
- Compare the 30 GB directly against the 48 GB of device memory, because the reader decompresses each row group in place and the loaded table therefore occupies the same bytes that were read from storage.
- Read one representative file, measure the in-memory footprint of the resulting table, and scale that by the number of files, because the stored bytes are compressed and encoded while the in-memory form is neither.