If you know your queries, however, you can do much more. You can check whether some of them would be executed more efficiently by a datastream system. Or by a NoSQL system. In both cases the queries must be translated into the languages understood by the resp. system.

And finally one can take queries also when the system needed to execute them does not even exist. One reason can be that the data needed to respond to the query are locally stored in your own system, but are available in external sources that still need to be attached. Another reason can be that there is no database system yet. Designing it could benefit substantially from knowing the queries that it is to execute later on. One can even try to generate parts of the database schema automatically from the queries.