Microsoft Certified: Azure AI Engineer AssociatePlan and manage an Azure AI solutionMedium
A retail company is developing an Azure AI solution to analyze customer sentiment from online reviews. They anticipate high traffic during promotional periods, requiring the sentiment analysis service to scale rapidly. The solution must also ensure data privacy by processing all data within a virtual network. Which Azure AI service deployment model best meets these requirements?
- AAzure AI services hosted in a multi-service resource
- BAzure AI services deployed as Docker containers on Azure Kubernetes Service (AKS)
- CAzure AI services deployed as serverless functions on Azure Functions
- DAzure AI services consumed directly via public endpoints
Show answer & explanationAnswer & explanation
Correct answer: B. Azure AI services deployed as Docker containers on Azure Kubernetes Service (AKS)
Deploying Azure AI services as Docker containers on Azure Kubernetes Service (AKS) allows for high scalability and the ability to run within a virtual network, addressing both high traffic and data privacy requirements. This model provides fine-grained control over the environment.
Why the other options are wrong
- A. Multi-service resources simplify management but don't inherently provide VNet isolation or the same level of scaling control as AKS containers.
- C. Azure Functions offer serverless scaling but typically run in a public cloud environment unless specifically configured with Premium plans and VNet integration, which adds complexity and may not offer the same level of control as AKS for containerized AI services.
- D. Consuming services directly via public endpoints does not meet the requirement for processing data within a virtual network, compromising data privacy.
Containerized Azure AI Services
Azure AI services can be deployed as Docker containers, allowing them to run on-premises or in Azure, providing portability, VNet isolation, and custom scaling.
- Enables deployment in private network environments.
- Offers control over data, security, and compliance.
- Supports disconnected environments and custom scaling.
Memory trick: Containers lock down AI, scaling global needs.