The Discovery Center performs key analysis automatically when you import data to a repository by creating a profiled data source. The analysis uses only a sample of your data and attempts to find the keys that meet default criteria for uniqueness.
There are two types of keys:
- Primary—Represents pieces of data (one or more attributes) that uniquely identify data and distinguish it from any other row within a data source.
- Composite—A primary key that is made up of more than one attribute (field or column).
How Key Analysis Works
After you import data, the Discovery Center presents a list of potential keys found during the import. The initial key analysis samples a maximum of 10,000 rows of data and tries to find potential single-attribute (primary) and double-attribute (composite) key combinations that are at least 98% unique.
The keys discovered may or may not be the keys that you expect in your data. Therefore, review the keys that are found and verify whether they are valid. If you believe that the list of discovered keys is incomplete, and that the sample data size is too small or the uniqueness percentage too high, create a discovered key with different criteria for the key analysis. Delete any discovered keys that are coincidental or irrelevant to your data profiling needs.