Data Tokenization
The replacement of sensitive data with non-sensitive tokens whose original values are stored or recoverable only through a controlled token system.
Tokenization can reduce the number of systems that store sensitive data such as card numbers or personal identifiers, which lowers breach exposure and can shrink compliance scope, for example under PCI DSS.
An online store sends card numbers straight to its payment provider, which returns a token. The store saves only the token for repeat purchases, so its order database and analytics tools never hold real card numbers.
Tokenization moves risk to the token vault or service, which must be secured and audited. Systems that can detokenize stay in compliance scope, and tokens derived from data with weak methods can sometimes be reversed or linked.
A tokenization system replaces a sensitive value with a surrogate, either by storing the mapping in a secured vault or by generating tokens cryptographically without a vault. Tokens can preserve format, such as length or last four digits, so existing systems accept them. Detokenization is restricted to authorized services. Unlike encryption, a vaulted token has no mathematical relationship to the original value.
Data Engineering, Systems Architecture
PCI Security Standards Council — Information Supplement: PCI DSS Tokenization Guidelines — https://listings.pcisecuritystandards.org/documents/Tokenization_Guidelines_Info_Supplement.pdf; Google Cloud Documentation — Pseudonymization — https://docs.cloud.google.com/sensitive-data-protection/docs/pseudonymization