Skip to content

Data Tokenization

Data Tokenization replaces a sensitive value with a token — a string with no mathematical relationship to the original value — and stores the encrypted original separately from the data source.

  • Vaulted Tokenization — Stores the encrypted original as a single entry in ALTR’s token vault.
  • Critical Tokenization — Splits the encrypted original across separate, isolated storage locations, for the highest-sensitivity values.

A token consists of a prefix and a globally unique identifier (GUID), for example vaultd_a1b2c3d4-e5f6-47a8-9b12-c3d4e5f6a7b8. The prefix identifies which tokenization variant and mode produced the token.

ALTR does not tokenize data automatically. Tokenization is available through several paths:

  • API. Tokenize values through the tokenization API, independent of Snowflake — data can be tokenized before it’s loaded into a data source.
  • Manual. Call the tokenization function directly against data already in a connected data source.
  • ETL-vendor integrations. Tokenize values as part of an ETL pipeline, using the tokenization function.
  • User-defined function (UDF). Tokenize values through a user-defined function on supported data sources.

Detokenization is available through several paths:

  • API. Detokenize values through the tokenization API.
  • Integration. Detokenize values as part of an integration, using the same API.
  • UDF. Detokenize values through a user-defined function, for users granted direct access to it.
  • Policy in Snowflake. Configure a masking policy with detokenization so authorized roles see the original value at query time. See Data Masking on Snowflake and Tokenization Access Policies.

A token has no mathematical relationship to the original value it replaces. Because the token is a random substitute rather than a derivation of the original, systems that store or process tokens are not handling the original sensitive data — this is intended to remove tokenized columns from Payment Card Industry Data Security Standard (PCI DSS) scope.

For Snowflake-specific tokenization mechanics — tag connections and native versus external-function masking — see Data Masking on Snowflake.