As mentioned earlier, data stored in Datastore is ultimately persisted in Bigtable. Every Datastore entity is actually persisted across six separate Bigtable tables, including a core entities table, three standard index tables, and two tables that support custom indexes.
How data is mapped to these six underlying Bigtable tables is directly responsible for the total cost of persisting data in Datastore, as well as the performance of all Datastore operations. Therefore, in order to architect data models to maximize performance and minimize cost, it is important to understand how this data is persisted in Bigtable.
