Allocate sufficient space in a local, NFS, or S3 storage for Parquet files.

Since Parquet files use a columnar format and are compressed, they are smaller than the equivalent unstructured text data or heap tables. The actual compression ratio depends on the selected compression algorithm and the order in which the data is written to Parquet files. For example, with the Snappy compression algorithm, Parquet files are 2 to 4 times smaller than the equivalent CSV data.

You can select a compression algorithm when configuring Parquet file storage parameters.

The storage throughput must be sufficient to support all analytical queries at peak workloads. For more information about estimating the required storage throughput, refer to Section 8.2.