...
DN lookups are more costly since more data is serialized back and forth. Plus there's a great amount of redundant information stored in the dn indices. If there are 1 million people under the ou=Users,ou=system entry then 1 million copies of ou=User,ou=system are started in the DN indices. With the new configuration there is no duplication, and the size of the data is minimized with just the Rdn.
| Note |
|---|
A related problem are attributes that store a distinguished names, for example the groupOfName's member attribute or the modifiersName attribute. When modifying the DN of an entry we also need to update all attributes of all entries that have a reference to the modified entry. A solution would to not store the DN of the referenced entry, but its ID or entryUUID. |