Protection des données sensibles

In this lesson, we'll learn how to protect sensitive data within your Erlang applications. But first, what exactly is sensitive data?

It's any information that, if exposed, could lead to harm, fraud, or privacy breaches. This includes:

• Personally Identifiable Information (PII) like names, addresses, or social security numbers.
• Financial data (credit card numbers, bank details).
• Authentication credentials (passwords, API keys).
• Proprietary business information.

Protecting sensitive data is crucial for several reasons:

• Trust: Customers and users expect their data to be safe.
• Compliance: Many regulations (GDPR, HIPAA) mandate strong data protection.
• Security: Prevents unauthorized access, data breaches, and financial losses.

We'll focus on protecting data at rest (stored), in memory, and how to manage encryption keys.

Data at rest refers to data stored on disk, in databases, or backups. To protect it, we use encryption, which transforms data into an unreadable format.

Erlang's built-in crypto module provides robust cryptographic functions. For data at rest, symmetric encryption is often used, where the same key encrypts and decrypts the data.

Let's see how to encrypt and decrypt a message using AES-256 in CBC mode, a common symmetric encryption algorithm. We'll need a key and an initialization vector (IV).

Try running this example:

Encryption is only as strong as its key. If an attacker gets your encryption key, they can decrypt your data. This leads to the critical question: Where do you store the encryption key itself?

• Never hardcode keys directly in your application code.
• Avoid storing keys alongside the encrypted data.

This is called key management, and it's one of the hardest parts of data security.

To protect your encryption keys, consider these approaches:

• Environment Variables: Load keys at application startup from environment variables, which are not stored in source control.
• OS-Level Secrets: Use operating system features (like `pass` on Linux or Windows Credential Manager).
• Hardware Security Modules (HSMs): Physical devices that securely store and manage cryptographic keys.
• Key Management Systems (KMS): Cloud-based services (AWS KMS, Azure Key Vault, Google Cloud KMS) designed for secure key lifecycle management.

Data in memory refers to sensitive information processed by your application (e.g., a user's password during login before hashing).

Erlang's process isolation helps, as each process has its own memory space. However, it's vital to:

• Minimize dwell time: Keep sensitive data in memory for the shortest possible duration.
• Clear memory: Explicitly overwrite or clear memory where sensitive data was stored, if possible (though Erlang's garbage collection handles much of this).

A common vulnerability is accidental exposure of sensitive data through logs or error messages.

• Never log sensitive data: Configure your logging system to filter out or mask sensitive information (e.g., credit card numbers, passwords).
• Sanitize inputs/outputs: Ensure that sensitive data is removed or obfuscated before being displayed to users, stored in non-secure locations, or sent to external services that don't need it.
• Secure crash dumps: Be cautious with crash dumps (`erl_crash.dump`) as they can contain process memory.

Effective data protection requires a multi-layered approach, combining various strategies:

• Encryption: For data at rest and in transit (using TLS, as covered in a previous lesson).
• Secure Key Management: Storing and handling keys with extreme care.
• Access Control: Limiting who can access sensitive data (both users and processes).
• Secure Coding Practices: Avoiding common pitfalls like logging sensitive data.
• Regular Audits: Periodically reviewing your security measures.

You've learned essential strategies for protecting sensitive data in Erlang applications:

• Identify Sensitive Data: Understand what needs protection.
• Encrypt at Rest: Use the `crypto` module for symmetric encryption.
• Secure Key Management: Never hardcode keys; use environment variables, KMS, or HSMs.
• Protect In-Memory Data: Minimize dwell time and prevent accidental logging.
• Prevent Leaks: Sanitize logs and outputs.

By applying these principles, you build more secure and trustworthy Erlang systems!