Use a unique index if possible. Otherwise, use a unique code.
In other words, if you can use a unique index, you should. If there is any reason why you need to record duplicate data, you have no choice yet.
If the index is guaranteed to be unique, Oracle can perform certain query plan optimizations that are not possible with an imperfect index β for example, if a unique index is used to determine a specific value, Oracle can stop processing further index blocks as soon as it finds a match.
source share