UNLOAD ('select-statement') TO 's3://object-path/name-prefix' IAM_ROLE '...' writes the result of a query to files on S3. FORMAT AS takes CSV, PARQUET or JSON, and with none the output is pipe delimited text. GZIP, BZIP2 and ZSTD compress it, HEADER writes column names, and MANIFEST VERBOSE writes a JSON file listing every output file with its row count and the totals. By default UNLOAD writes one or more files per slice; PARALLEL OFF writes serially in ORDER BY order. Maximum file size defaults to 6.2 GB and MAXFILESIZE sets it anywhere from 5 MB to 6.2 GB. The user needs SELECT on the data, permission to assume the IAM role, and access to the bucket. The outer SELECT cannot carry a LIMIT clause, so nest it.
What it holds
Warehouse tables and their rows, and materialized views.
Find the size
SVV_TABLE_INFO holds one row per user table and materialized view in the connected database: size is the table size in 1 MB data blocks, and tbl_rows is the total row count including rows marked for deletion but not yet vacuumed, with estimated_visible_rows excluding those. It is visible to superusers only unless somebody grants SELECT on it, and it returns nothing for an empty table.
What to enter on the form
Amazon Redshift in GB, and the year your company started using it. Seller form