Redshift Spectrum Cost Considerations. Charges for Spectrum are calculated based on the number of bytes scanned rounded up to the next megabyte. There is a 10-megabyte minimum per query. There are no charges for Data Definition Language, DDL, statements. So you can run as many CREATE, ALTER, and DROP TABLE statements as needed. In general, and I've mentioned this in other ways, query performance can be improved, and costs can be reduced, by storing data in a compressed, partitioned, and columnar format.

If data is compressed using one of Spectrum's supported formats, cost go down because less data is scanned. Similarly, if data is stored in a columnar format such as Apache Parquet or Apache ORC, there will be fewer changes because Spectrum only scans columns needed by the query. As a reminder, Spectrum is part of Amazon Redshift, so you will be charged for any Redshift clusters provisioned. Data stored in S3 also include charges based on the amount of data stored as well as the number of requests made against S3 buckets. If you use the AWS Glue Data Catalog with Redshift Spectrum, you will be charged standard Glue Data Catalog rates. Finally, if you use KMS to encrypt data inside S3 buckets, you will be charged standard AWS KMS rates.