An advertising technology company hosts a real-time bidding application on Azure virtual m… — Microsoft Azure Solutions Architect (AZ-305)

An advertising technology company hosts a real-time bidding application on Azure virtual machines (VMs) in the East US region. The application architecture consists of three front-end web server VMs and one PostgreSQL database VM. The database VM has a single Premium SSD data disk with a continuous write churn of $25\text{ MB/s}$ . The front-end VMs have minimal disk write activity. You are designing a disaster recovery solution to the West US region. The solution must meet a Recovery Point Objective (RPO) of $15\text{ minutes}$ and a Recovery Time Objective (RTO) of $1\text{ hour}$ . Which replication strategy should you recommend to meet the requirements?

Replicate the front-end VMs using Azure Site Recovery. Replicate the database VM using PostgreSQL native streaming replication to a target VM in the West US region.Answer
B
Replicate all VMs, including the database VM, using Azure Site Recovery with the High Churn replication policy.
C
Replicate all VMs using Azure Site Recovery, and exclude the database VM's Premium SSD data disk from replication.
D
Use Azure Backup to back up the database VM to a Geo-Redundant Recovery Services vault, and replicate the front-end VMs using Azure Site Recovery.

Answer

Replicate the front-end VMs using Azure Site Recovery, and replicate the database VM using PostgreSQL native streaming replication to a target VM in the West US region.

The correct strategy is to replicate the front-end VMs using Azure Site Recovery and replicate the database VM using PostgreSQL native streaming replication. This is because Azure Site Recovery limits block-level write replication churn to a maximum of

20\text{ MB/s}

per Premium SSD disk under the High Churn configuration. Because the database disk exhibits a continuous write churn of

25\text{ MB/s}

, utilizing Azure Site Recovery for the database VM is unsupported. Native streaming replication handles the database replication, while Azure Site Recovery is perfectly suited to meet the

15\text{ minute}

RPO and

1\text{ hour}

RTO for the low-churn front-end VMs.

Step-by-Step Solution

Analyze the write churn requirements of the workload components against Azure Site Recovery limits.

The database VM disk write churn is

25\text{ MB/s}

, which exceeds the maximum supported threshold of

20\text{ MB/s}

per Premium SSD disk for Azure Site Recovery (even with the High Churn policy enabled).

This establishes that the database VM cannot be replicated using Azure Site Recovery.

Evaluate replication alternatives for the database VM that can meet the

15\text{ minute}

RPO and

1\text{ hour}

RTO.

PostgreSQL native streaming replication can continuously replicate transaction logs to a target VM in the secondary region, meeting both the RPO and RTO constraints.

Database-level replication bypasses disk-level block replication limits imposed by virtual machine replication tools.

Determine the optimal disaster recovery mechanism for the front-end web server VMs.

Replicate the front-end VMs using Azure Site Recovery, which easily handles their low churn and meets the low RPO/RTO goals.

Using Azure Site Recovery for the application tier ensures rapid failover orchestration and matches the target RTO.

Key Concept

Azure Site Recovery limits and hybrid disaster recovery design

Question

Answer

Step-by-Step Solution

Key Concept

Common Mistakes

Topics

More questions from this topic