- insertion anomalies
- You may want to add information about a person with whom you want to do business. The above table only allows information for customers that own a share of stock. If the person does not own stock, then the last four columns in the table have to be empty. This is not allowed since stock is part of the primary identifier.
- deletion anomalies
- You may want to delete a record from the above table because a customer sold his stock. Consider the second record. If this record were deleted, then information would be lost about both Jones's address and the price of C stock.
- update anomalies
- You may want to update a customer's address, the price of a stock, or its most recent dividend. To accomplish this update, you would have to update several rows in the table. If you miss one of the rows that should be updated, then at a later time you will get two different answers to a question you ask of the data. This is not good. Generally, if there is only supposed to be one answer to a question, you want to get just that one answer. (I realize this is not a deep insight on my part.)
Back to discussion about normalization.
page revision: 1, last edited: 27 Jul 2008 19:20