Search code examples
databasegoogle-fusion-tables

Data Normalization in Fusion Tables

I'm about to start building an application on top of Google's Fusion Tables service and have a simple schema:

| date | vehicle | location |
-----------------------------
| 2001 | Car     | 37,-78   |
| 2002 | Truck   | 36,-50   |

I have a column, vehicle, that I could either put an integer to represent the vehicle type, or put in the text "Car", "Truck", etc... This would make the table more human readable, but might hurt performance when I filter on the vehicle column.

Does anyone know if Google attempts to normalize the text data for me or whether filtering on text fields does hurt performance?

Should I create a separate vehicle table and a view that joins the two?

Thanks!

Solution

  • In database systems, normalize doesn't generally mean "replace text with an id number". It would be pretty surprising if Google did that on the fly. Google generally follows the Principle of Least Surprise.

    Not only would it be surprising, it might require a join for every text column. That doesn't sound wise, does it?