AZ-305 Question 170
Select 2Your company is building a globally distributed e-commerce application that must handle high read and write volumes for users in multiple geographic regions. The application follows a microservices architecture, and the requirement is to minimize latency for both reads and writes while ensuring high availability. Additionally, front-end traffic should be directed intelligently to the nearest available region. Which of the following Azure design considerations should be incorporated into the architecture? (Choose two.)
- A
Use Azure Cosmos DB configured with multi-region writes to ensure low-latency data operations globally.
- B
Deploy a single Azure SQL Database in the primary region and direct all read/write traffic there for consistency.
- C
Implement stateful microservices on Azure Kubernetes Service (AKS) using ephemeral disk for persistent data storage.
- D
Establish an active-active deployment across multiple Azure regions behind Azure Front Door for load distribution.
- E
Store session state in an unmanaged virtual machine file share replicated across regions.
Show answer and explanation
Correct answers: A, D
Explanation
Designing an application architecture for global scale and high availability often involves distributing both computing resources and data storage across multiple Azure regions. Using Azure Cosmos DB with multi-region writes (Option 1) provides globally distributed, low-latency data operations for read and write workloads. Complementing this with an active-active deployment (Option 4) behind Azure Front Door ensures user requests are routed efficiently to the closest available region for optimal performance. For more details, refer to Azure Architecture Center guidelines on global web application design: https://learn.microsoft.com/azure/architecture/best-practices/global-web-app.
- A. Correct.
Option 1 is CORRECT. Azure Cosmos DB with multi-region writes enables high availability and low-latency data access for both reads and writes. It supports multi-master replication, which is particularly useful for globally distributed applications that need fast, local data access in each region.
- B. Incorrect.
Option 2 is INCORRECT. Relying on a single SQL Database in one region forces all traffic to that region, increasing latency for users distant from the primary region. While it may maintain consistency, it doesn't fulfill the requirement of minimal latency for global users.
- C. Incorrect.
Option 3 is INCORRECT. Storing persistent state in ephemeral disks on AKS is not recommended. Ephemeral disks do not survive pod restarts or re-deployments. For reliability and consistency, a managed data store like Azure Cosmos DB, Azure SQL Database, or Azure Managed Disks is more appropriate.
- D. Correct.
Option 4 is CORRECT. An active-active strategy that deploys microservices across multiple Azure regions and leverages Azure Front Door to route traffic ensures minimal latency and high availability. Each request is routed to the nearest, healthy back-end region.
- E. Incorrect.
Option 5 is INCORRECT. Storing session state in an unmanaged file share replicated across regions is cumbersome, less scalable, and can introduce high latency. Managed services (e.g., Azure Cache for Redis) or other distributed caching/store solutions are more suitable for session data.