Скачать или смотреть Database Normalization 1: First Normal Form and Data Anomalies

In this video we begin a discussion of database normalization. In normalization, we take more of a bottom-up approach. We start with some sample data. Our first rule is that each uniquely identifiable "cell" (specifying the table, the column and the row) should have a single-value piece of data. No lists. If this is not the case, then we "flatten" the file by creating a row corresponding to each value in the multi-valued field. This process results in repeating the info that was not multi-valued hence introducing redundancy. We talk about various data anomalies: insert anomaly, delete anomaly, and update anomaly. Splitting the table into multiple tables will reduce the redundancy but we have to make sure that we do no lose any information (relationship) along the way.