บัญชีดำและบัญชีขาวของ JWT

JSON Web Tokens (JWTs) are powerful for authentication, but sometimes you need to invalidate them before their natural expiry. This process is called token revocation.

• Compromised Token: If a token is stolen.
• User Logout: To immediately end a user's session.
• Password Change: To invalidate all old tokens.
• Role Changes: To force re-authentication with new permissions.

JWTs are inherently stateless. Once issued, they contain all necessary information for validation and don't require the server to store session data.

This statelessness is a strength, but it makes direct server-side revocation tricky. The server typically doesn't hold a list of active tokens to simply 'turn off'.

Blacklisting is a common strategy to revoke JWTs. When a token needs to be invalidated, its unique identifier (often the JTI claim) is added to a 'blacklist' — a list of forbidden tokens.

• Any token whose JTI is on this list is rejected, even if it's cryptographically valid and not expired.
• This allows you to 'undo' a token's validity.

The blacklist needs to be stored in a highly available, fast-access data store. Speed is crucial because every incoming request might need to check this list.

• Redis: An excellent choice due to its in-memory nature and support for time-to-live (TTL) on entries, which can match token expiry.
• Database: A simple table can work, but might be slower for high-volume checks.
• Each entry typically stores the JWT's JTI and its original expiry time.

Here's a basic interface for a service that manages a token blacklist. In a real application, this would interact with a database or a caching system like Redis.

When a request arrives, a security filter would first validate the JWT's signature and expiry. Then, it would check if the token's JTI is present on the blacklist before granting access.

Try running this example:

Whitelisting is an alternative revocation strategy. Instead of listing forbidden tokens, you maintain a list of active, allowed tokens.

• When a token is issued, its JTI is added to a 'whitelist'.
• For every request, the token's JTI must be found on this whitelist to be considered valid.
• If a token's JTI is not on the whitelist, it's rejected.

Similar to blacklisting, a whitelist requires a fast, persistent store (e.g., Redis). The key difference is what you store and how you manage it:

• Each entry typically stores the JWT's JTI, often associated with a user ID.
• When a user logs out or changes their password, all active JTIs associated with that user can be efficiently removed from the whitelist.

Both strategies achieve revocation but have different implications:

• Blacklist: Ideal for rare, specific revocations (e.g., single token compromise). Requires less storage if revocations are infrequent.
• Whitelist: Better for frequent revocations (e.g., user logout invalidates all tokens). Can simplify session management but requires more storage for all active tokens.
• The choice depends on your application's specific needs and the frequency of revocations.

Consider an application where users frequently log out, and you need to ensure all their issued tokens are immediately invalidated upon logout.

In this lesson, we've explored advanced strategies for revoking JWTs, which is crucial for robust security:

• Blacklisting: Marking specific tokens as invalid by adding their JTI to a forbidden list.
• Whitelisting: Only allowing tokens that are explicitly listed as active, often tied to a user session.
• The best approach depends on your application's requirements, especially the frequency and nature of token invalidation.

Next, we'll analyze the performance implications of these techniques.