
Firstly, the grouping of attributes in version B is less logical than in version A. The
latter version seems intuitively more sensible because there is one table which contains
attributes of customers only, while the other table relates orders to customers. In
version B, both tables relate orders to customers.
Secondly, version B can contain redundant data. Suppose the value Smith were
deleted from row 1 of Customer2. The missing value could be deduced by using the
orderNo (1) to find the corresponding customerNo in OrderCust (C1), then searching
OrderCust for another orderNo placed by C1 (orderNo 3 or 4), and finally using one of
these orderNo(s) to ...