Question

A retail analytics team runs Amazon Redshift for their data warehouse. The most recent 12 months of sales data is queried frequently and must return results with low latency. However, they must retain 7 years of historical sales data for occasional compliance and trend-analysis queries. Storing all 7 years in Redshift managed storage has become expensive, and most of the older data is queried only a few times per month. The team wants to reduce warehouse storage costs while keeping all historical data queryable through their existing Redshift SQL interface. What is the MOST cost-effective solution?

Accepted Answer

Keep the recent 12 months in Redshift local tables, unload the older data to Parquet files in Amazon S3, and query it via Redshift Spectrum using external tables in the Glue Data Catalog.

Answer

Migrate all 7 years of data into a single DynamoDB table and use Redshift federated queries to access the historical partitions.

Answer

Increase the Redshift cluster node count and enable concurrency scaling so all 7 years remain in managed storage with acceptable performance.

Answer

Export the older data to Amazon S3 Glacier Deep Archive and restore it into Redshift on demand whenever a compliance query is required.