AZ-305 Question 134
Single answerContoso plans to migrate a mission-critical web application to Azure with a requirement of 99.99% availability in a region that supports availability zones. The application consists of multiple front-end virtual machines that handle user requests and must remain online even during planned maintenance or localized Azure infrastructure failures. Which high availability approach should you recommend?
- A
Deploy a single Virtual Machine (VM) on a Premium SSD with an Azure Backup job to ensure uptime.
- B
Use an Availability Set with two or more VMs distributed across multiple fault and update domains.
- C
Distribute two or more VMs across multiple availability zones behind a zone-redundant load balancer.
- D
Implement a single-instance Virtual Machine Scale Set (VMSS) configured with a custom auto-scaling policy.
Show answer and explanation
Correct answer: C
Explanation
To achieve a 99.99% SLA for compute resources in an Azure region supporting availability zones, you should deploy multiple VMs across at least two availability zones and place them behind a zone-redundant load balancer. This architecture ensures redundancy at the data center level and can withstand infrastructure failures or planned maintenance in any single zone. For more details, refer to the official Microsoft documentation: https://learn.microsoft.com/azure/virtual-machines/availability#availability-zones.
- A. Incorrect.
Option 1 is incorrect: A single VM using Premium SSD might offer a 99.9% SLA, but it doesn’t meet the 99.99% requirement. Backup jobs help with data recovery, not continuous uptime.
- B. Incorrect.
Option 2 is incorrect: An Availability Set helps achieve around 99.95% SLA by distributing VMs across fault and update domains, but it does not fulfill the 99.99% requirement that availability zones can provide.
- C. Correct.
Option 3 is correct: Using multiple VMs spread across availability zones behind a zone-redundant load balancer provides resiliency across physically separate datacenters in the same region, supporting a 99.99% SLA.
- D. Incorrect.
Option 4 is incorrect: A single-instance Virtual Machine Scale Set does not distribute workloads across multiple VMs. Even with custom auto-scaling, it won’t offer higher availability on its own because there is no redundancy across zones.