The following limits apply when you embed watermarks in data sources.
What Data Sources Are Suitable for Watermarking?
Data Security Center supports embedding watermarks in all data sources that support static data masking. For a list of data source types that support static data masking, see Supported asset types.
What are the source data requirements for watermark embedding?
- The source data must have 1000 or more rows.
For source data with fewer than 1000 rows, watermark extraction may fail because of insufficient features.
- Embed watermarks in columns with diverse data values. If the column values are fully enumerable, extraction may fail because of insufficient features.
Columns suitable for watermarking include addresses, names, UUIDs, amounts, and totals. Columns not suitable for watermarking include gender and status.
- Watermarked data will differ from the raw data. Therefore, insert watermarks only into fields that can tolerate value changes.
For more information about the differences, see Does embedding a watermark modify raw data?.
Choosing a watermark algorithm
Select the space algorithm for character-type fields and the modify least significant bit algorithm for numeric-type fields.
Does embedding a watermark modify raw data?
When you configure a task in the Data Security Center console to embed a watermark in data from Table A and write it to Table B, the data in Table A is not modified. Only the data written to Table B is affected.
The values of some watermarked data will differ from the raw data values. The supported algorithms are the space algorithm and the modify least significant bit algorithm.