Columns related as being likely to be joined in a database query can be identified based upon primary and foreign key constraints such as those defined by an administrator, based upon naming structure such as each column having the same name, or based upon the access plan cache of completed queries.