Yeah, but a data warehouse isn't supposed to have great response times. Data warehouses are for large, low-value sets of historical data that you don't always know how you want to use.

If you want to use data in real-time, you should be driving it from your transactional systems. Redshift and other data warehouse solutions are for doing reporting and dashboards, not triggering real-time reactions.

Well, used to be true, but now those systems are converging. -- Full disclosure, I work for a company working on exactly that problem called Treasure Data.

Most companies are generally more concerned about reducing their data warehouse costs than they are about improving the performance of their data warehouses. Many companies implement a multi-tiered DW structure to get a mix of the two, but the core driver is managing the cost of storing petabytes of data while keeping performance acceptable.