Databricks Generative AI Engineer Associate Question 190
Select 3You are deploying a machine learning model on Databricks Model Serving and need to control access to its serving endpoint. Which actions can you take to restrict and secure access to the endpoint?
- A
Configure access control lists (ACLs) for the serving endpoint to define who can query it.
- B
Use a token-based authentication mechanism to enforce secure access.
- C
Allow unrestricted public access to the serving endpoint for better scalability.
- D
Integrate the serving endpoint with an IP allowlist to restrict access to specific networks.
- E
Disable role-based access control (RBAC) for the serving endpoint to simplify permissions.
Show answer and explanation
Correct answers: A, B, D
Explanation
To secure access to Databricks Model Serving endpoints, you should use a combination of access control lists (ACLs), token-based authentication, and network-level restrictions like IP allowlisting. These measures ensure that only authorized users and traffic are allowed, while unrestricted public access or disabling RBAC would compromise the endpoint's security.
- A. Correct.
Correct: Configuring ACLs allows you to define fine-grained permissions for users or groups, ensuring only authorized users can access the endpoint.
- B. Correct.
Correct: Token-based authentication is a secure method to validate requests made to the endpoint, ensuring that unauthorized users cannot access it.
- C. Incorrect.
Incorrect: Allowing unrestricted public access exposes the endpoint to potential abuse and security risks, making it unsuitable for secure environments.
- D. Correct.
Correct: IP allowlisting restricts traffic to the endpoint from specific IP ranges, adding an extra layer of network-level security.
- E. Incorrect.
Incorrect: Disabling RBAC removes a critical layer of security, leading to unauthorized access and lack of control over who can perform specific actions.