Supplementary — not from your PDF Intermediate ~40 min

Mask and tokenize data in a database

With Python's built-in SQLite, store fake customer data, give support staff a masked view, and replace card numbers with tokens held in a separate vault table.

Environment

Python 3 on your own computer (the sqlite3 module is built in).

Before you start

Read Obfuscation (p.78), Database Encryption (p.72) and Data Protection (p.428).

You will

Create a masked view
Implement simple tokenization
Compare masking, tokenization and encryption

Steps

1
In Python: import sqlite3, secrets; db = sqlite3.connect('shop.db').
2
Create a table and add fake rows: db.execute('CREATE TABLE customers (id INTEGER PRIMARY KEY, name TEXT, phone TEXT, card TEXT)'), then insert three made-up customers (use the test card number 4111111111111111).
3
Create a masked view for support staff: db.execute("CREATE VIEW support_view AS SELECT id, name, substr(phone,1,3) || '-xxx-xxxx' AS phone, '**** **** **** ' || substr(card,-4) AS card FROM customers") and query it.
4
Tokenize: create vault(token TEXT PRIMARY KEY, card TEXT). For each customer, generate token = secrets.token_hex(8), store (token, card) in the vault, and replace the card column with the token.
5
Show that the customers table no longer holds card numbers, and that an authorized lookup in the vault gets them back.
6
Write a table comparing masking, tokenization and encryption: reversible? keys needed? where the real data lives?

Check your understanding

?Why is tokenization often used for card data in scope for compliance?
?Is masking reversible for the person viewing the masked data?
?What becomes the most important asset to protect once you tokenize?