In a typical/traditional data warehouse solution, the data is read into ETL memory, processed/transformed in the memory before loading into the target database. With the growing data, the cost of compute is also increasing and hence it becomes vital to look for an alternate design.
Welcome to pushdown query processing. The basic idea of pushdown is that certain parts of SQL queries or the transformation logic can be "Pushed" to where the data resides in the form of generated SQL statements. So instead of bringing the data to processing logic, we take the logic to where data resides. This is very important for performance reasons.
