Let me summarize how interaction with SQL works in DSS. You have very different kinds of things:
* The DATASET : it gets the rows of a SQL table, or the results of a SQL query, and makes them available as a dataset, ie, as a list of rows that you can see, process, export, visualize, ...
* The RECIPE: the "SQL Query" recipe is used to create a new dataset (stored anywhere), as the results of a SQL Query. The target dataset can be either a SQL table, or just any kind of dataset you want.
So, a DATASET is more like access on existing data, while one uses a RECIPE to create a new dataset from existing datasets, in a reproducible Flow.
More information about the SQL datasets: http://doc.dataiku.com/dss/latest/connecting/sql_datasets.html
More information about the SQL recipes: http://doc.dataiku.com/dss/latest/code_recipes/sql.html
The most important concept is that in both the Dataset and Recipe, it only makes sense to make SELECTs. The whole idea is to generate rows.
Now there is a third thing which is the SQL NOTEBOOK: it's simply an interactive SQL evaluation UI, where you can pass any SQL statement, and it gets executed.
If you want to create a SQL table, you can do it directly in the Notebook, but it will not be specially known by DSS, and it won't help you perform operations on it.
I would kindly suggest that you go through the tutorials that give more insight about how exactly datasets and recipes interact.