Do dynamic tables need a primary/logical key defined in the underlying tables?

No, Dynamic tables in Snowflake don't inherently require a primary or logical key defined in the underlying tables they reference.

Here's why:

Dynamic tables are virtual representations of data retrieved from other tables or views. They don't store data themselves.
Joins and filtering within the dynamic table definition determine which rows are included in the result set. These can leverage columns from the underlying tables that act as unique identifiers even without a formal primary key.
However, there are situations where having a primary or logical key in the underlying tables can benefit dynamic tables:

Performance: If you frequently filter the dynamic table based on a specific column, having that column defined as a primary or unique key in the underlying table can improve query performance.
Data Integrity: Primary keys help ensure data consistency, especially during updates or deletes in the underlying tables. While dynamic tables themselves don't enforce these constraints, the underlying tables with primary keys might.
In summary, while not mandatory, defining primary or logical keys in the underlying tables referenced by a dynamic table can enhance performance and data integrity in certain scenarios.