Finding and exporting your Amazon Redshift data

What it holds

Redshift stores analytical business tables and event histories. Table rows represent the retained warehouse contents, while external datasets may live separately in object storage.

Where Amazon Redshift shows the size

1. Connect as a superuser, or ask one to run GRANT SELECT ON svv_table_info TO username;
2. In each database run SELECT "schema", SUM(size) AS size_mb, SUM(estimated_visible_rows) AS visible_rows, MIN(create_time) AS first_table_created FROM svv_table_info GROUP BY "schema" ORDER BY size_mb DESC;
3. Divide size_mb by 1024 for binary GB. Use CloudWatch PercentageDiskSpaceUsed as a cluster-level cross-check.

Plan and admin role. SVV_TABLE_INFO requires superuser access unless SELECT has been delegated. UNLOAD requires data access and an associated IAM role authorized to write to the destination S3 bucket.

Exporting the full history

1. Associate the S3-writing IAM role with the cluster, optionally as its default role.
2. Run UNLOAD ('SELECT * FROM public.orders') TO 's3://my-bucket/export/orders/' IAM_ROLE default FORMAT AS PARQUET MANIFEST VERBOSE;
3. Add REGION 'us-west-2' for a bucket in a different Region. Repeat for other tables and inspect the verbose manifest's file and row counts.

SVV_TABLE_INFO omits empty tables and covers only the connected database.

UNLOAD normally writes parallel files up to 6.2 GB each and uses SSE-S3 encryption by default. Parquet uses Snappy compression.

TIMESTAMPTZ loses time-zone information in Parquet. GEOMETRY, HLLSKETCH and VARBYTE require text or CSV output; the outer SELECT cannot contain LIMIT.

You don't need to run an export to talk to us, and there's no need to clean it first — removing personal data is part of our work.