How to extend Delphix Profiler

Below is an excerpt from a very common and useful conversation;
Question
Hi I have a question about profiling. I was playing around in Masking and I chose the HIPAA profile set. Since the column names are not exactly FIRST_NAME, LAST_NAME, it couldn’t recognize the algorithm automatically. I thought the profile is smart enough to see a name or first and/or any combination of that and automatically assign the FIRST_NAME as the algorithm?

Is it not true? Please let me know. Thanks.

Answer:

Hi,

The Delphix discovery framework consists of 2 major artifacts

• Profiler Expressions: Regular expression used to match an identifier with the intended sensitive data element pattern. This can be created both for metadata and data.
• Profiler Sets: Groups of Profiler expressions, this group/set/pack is used by the profile job on a target masking rule. A single sensitive element may correspond to multiple regular-expression for more coverage additionally can be both column and data level.

 

In order to scan a data type in a column named differently than the original HIPPA one simply follow the below steps

 

• Step1:  Add a new expression ( e.g FIRST_NAME_ALTERNTE)

Here in the e.g (?i) is for case insensitive and | separated values are optional column names (FIRST_NAME|FST_NAME|FIRST_NOM|FST_NM). You can be more creative by using the advanced regex patterns for more greedy or fine POSIX regex. try here: https://www.regextester.com/

• Step 2: Add the newly created Expression (FIRST_NAME_ALTERNTE) to the Profiler set you to plan to use (or you can create a new one).

• Step 3: Run profiling job with updated profiler set.

You can keep adding more expressions or tweak them to your choice of the result. You may need a few iterations to have a result of your liking. Please note Column level expressions run instantaneously while the Data level will actually read data from the table and scan so will run per speed, memory, etc of the database.

Please let me know if you have any more questions.

Thanks

Hims

--