Best (most efficient) DataType to use for UUIDs as JPA IDs

Question

I want to use UUIDs as IDs for my JPA Objects.

I am currently just using a String to store the UUID. What would be more efficient?

Answer 1

How are you measuring efficiency?

For example, storing the UUID (which is a text encoding of a byte[]) as a couple of long values will allow you to compare them very quickly on a 64-bit architecture (much more quickly than String comparison, which is character-by-character). However, your coding efficiency would suffer, because you have to write a custom type.

What is important in your case?

---

If you are interested in performance within the database, performance will depend somewhat on the database you choose, but all will essentially compare UUIDs byte-by-byte. So, the important thing is to make differences between keys more likely to be toward the beginning of the key.

UUIDs are well-designed in this respect. The best UUID is the "random" style. They don't have a lot of common substrings anywhere, including the beginning, and they don't leak information about the machine that generated them. But, the "timestamp" style of UUIDs makes a good key as well, since the low-order bits (that vary quickly) are first in the string, while the high-order bits (that don't change for successive UUIDs) come later.