I am unable to get that what advantage do we get when we store only records consisting of primary attributes in case of B+ tree i.e. we do not store record pointers corresponding to those key attributes , we do it only at the leaf level , doesn't it increase the overhead since we have to come upto the leaf level and then access the data where as in case of B tree with each node we have a record pointer associated so I guess that's more advantageous , so then why database designers prefer B+ trees over B trees ?