You're comparing redshift (OLAP) with Athena (not OLAP but a federated query engine). PrestoDB/Athena is good but it's not an OLAP database. There are OLAP sub-second databases that now can query other JDBC sources, Apache Iceberg, Apache Hudi and other files in S3. There is movement now to move to an open data lake architecture with Apache Iceberg and Apache Hudi with a query engine like StarRocks to aggregate data from those open table formats and JDBC resources to replace the use case you describe.