AI-102 Question 456
Select 3You are designing an Azure-based application that uses the DALL-E model to generate images based on user-provided text prompts. The application must support high availability, secure access to the DALL-E API, and ensure cost management. Which actions should you take to meet these requirements?
- A
Use Azure OpenAI Service to access the DALL-E model and implement Azure Active Directory (AAD) for secure authentication.
- B
Deploy the DALL-E model on a dedicated Azure Kubernetes Service (AKS) cluster for high availability.
- C
Set up a usage quota and cost alert for the Azure OpenAI Service to monitor and control spending.
- D
Enable a private endpoint for the Azure OpenAI Service to ensure secure access to the DALL-E API.
- E
Leverage Azure Load Balancer to distribute traffic to multiple deployments of the DALL-E model.
Show answer and explanation
Correct answers: A, C, D
Explanation
To meet the requirements, leveraging Azure OpenAI Service provides managed access to the DALL-E model and ensures high availability. Implementing Azure Active Directory secures authentication, while usage quotas and cost alerts help manage spending. Adding a private endpoint enhances security by restricting access to a private network. Deploying the model on AKS or using an Azure Load Balancer is unnecessary because Azure OpenAI Service already manages these aspects.
- A. Correct.
Using Azure OpenAI Service provides managed access to the DALL-E model, and integrating Azure Active Directory ensures secure authentication and role-based access control.
- B. Incorrect.
Deploying the model on Azure Kubernetes Service is unnecessary because the DALL-E model is already hosted and managed by Azure OpenAI Service.
- C. Correct.
Setting up a usage quota and cost alert is a best practice to monitor and control expenses when using Azure OpenAI Service.
- D. Correct.
Enabling a private endpoint ensures that access to the Azure OpenAI Service is secure by communicating over a private network, reducing exposure to the public internet.
- E. Incorrect.
Azure Load Balancer is not applicable in this scenario because the DALL-E model is managed as part of the Azure OpenAI Service, which handles scaling and availability.