The flip side is that GUIDs take up a ton of space and are full of entropy, which makes them very inefficient to work with, are not cache-friendly, and not sortable. Certain database engines (e.g., InnoDB on MySQL) also limit the length of primary keys for performance reasons, which disqualifies GUIDs from being used. With clustering, short keys are better as they reduce IOPS required to fetch data.

See https://www.percona.com/blog/2006/10/03/long-primary-key-for... for an interesting discussion of the topic.

In general, it's unwise to make recommendations about what data types to use, especially for primary keys, without thorough research on their implications.