The Professional Cloud Architect exam expects you to recognize the relational concepts that BigQuery now supports, even though BigQuery started life as a non-relational analytics warehouse. The model itself is straightforward, but the exam tests whether you can name the parts and explain why referential integrity matters.
A relational database organizes data into tables, where each table holds rows about a single entity. A typical example uses three tables: a customers table, an orders table, and a products table. Each table represents one kind of real-world thing, and the database connects them through keys.
The whole point of the relational model is that data should be split across these entity tables and then related back together using keys. That separation keeps the data organized and coherent. Customer details live in one place. Order details live in another. The database stitches them together when you query.
A primary key is a column, or a set of columns, that uniquely identifies each row in a table. In the customers table the primary key is customer_id. In the orders table it is order_id. In the products table it is product_id.
That uniqueness guarantee is what makes the table usable. If I want to find Mike Smith, I look up customer_id = 151 and I know I get exactly one row back. If I want to delete Mike Smith because he requested deletion, I can find that row and any rows in other tables that reference customer_id = 151 and remove them cleanly. Without a primary key you cannot reliably point at a single row.
For the Professional Cloud Architect exam you should be able to state this definition in one sentence: a primary key is a column or set of columns whose values are unique across the table.
A foreign key is a column in one table that refers to the primary key in another table. The foreign key is the mechanism that establishes the relationship between two tables.
In the orders table, customer_id is a foreign key. It points back to customer_id in the customers table. Each row in the orders table is linked to a row in the customers table through that column. The orders row is the child. The matching customers row is the parent. One customer can have many orders, but every order has exactly one customer.
The foreign key is just data. It is a column with values in it. What makes it a foreign key is the declared relationship that says these values must correspond to primary key values in another table.
Referential integrity is the principle that foreign key values must always correspond to valid entries in the parent table. If the orders table has a row with customer_id = 101, then the customers table must contain a row with customer_id = 101. If it does not, that orders row is an orphaned record. It is a child with no parent.
Orphaned records break the model. Reports go wrong. Joins drop rows. Customer-facing tools display rows that point at nothing. A relational database that enforces referential integrity rejects any insert or update that would create an orphan, and it rejects any delete on the parent that would leave existing children stranded.
This is the concept the Professional Cloud Architect exam is most likely to ask about by name. If you see "referential integrity" in a question, the answer involves foreign key values matching primary key values in the parent table.
BigQuery historically did not enforce primary keys or foreign keys at all. It was a pure analytics engine where you joined on columns by convention. More recently BigQuery added unenforced primary key and foreign key constraints, which means you can declare them in a table schema and the optimizer will use them, but BigQuery does not block writes that violate them. The data architect is still responsible for keeping the data clean.
For the exam, the takeaway is that you should know the vocabulary. Primary key, foreign key, referential integrity, parent, child, orphaned record. These are foundational data architecture terms and the Professional Cloud Architect exam treats them as background knowledge a cloud architect is expected to have.
My Professional Cloud Architect course covers the relational model alongside the rest of the storage and analytics material.
