When might the use of a primary key in a database be undesirable or impractical?

Applying a primary key in a database may be undesirable or impractical in the following cases:

• Temporary or log file tables — when data is used temporarily and does not require a unique identification.
• Tables with duplicate records, where uniqueness is not important, such as an event log without the need for a unique identifier.
• Very large tables with high write load, where maintaining a primary key index can reduce performance.
• Cases where data is aggregated or used as a cache, and record uniqueness is not critical.

However, even in such cases, it is generally recommended to have at least one unique identifier for ease of data handling and ensuring integrity.