Google Cloud: Designing High Availability for Compute Engine Virtual Machines in us-central1

Designing High Availability for Compute Engine Virtual Machines in us-central1

Question

You are hosting an application from Compute Engine virtual machines (VMs) in us"central1"a.

You want to adjust your design to support the failure of a single Compute Engine zone, eliminate downtime, and minimize cost.

What should you do?

Answers

Explanations

Click on the arrows to vote for the correct answer

A. B. C. D.

A.

The best option to adjust the design and support the failure of a single Compute Engine zone, eliminate downtime, and minimize cost is Option A: Create Compute Engine resources in us"central1"b. Balance the load across both us"central1"a and us"central1"b.

Explanation:

In Google Cloud, Compute Engine VMs are organized into zones, which are distinct data centers within a region. By creating Compute Engine resources in both us"central1"a and us"central1"b, we can distribute the workload and prevent a single point of failure. If one zone fails, the other zone can handle the workload without downtime.

To balance the load across both us"central1"a and us"central1"b, we can use a load balancer. A load balancer distributes incoming traffic across multiple Compute Engine instances in multiple zones. In this case, we would want to use a regional load balancer that distributes traffic across both us"central1"a and us"central1"b.

Option B, creating a Managed Instance Group and configuring the Health Check with a short Health Interval, does not eliminate downtime or prevent a single point of failure. If the entire zone fails, the Managed Instance Group would not be able to handle the workload.

Option C, creating an HTTP(S) Load Balancer with global forwarding rules, would distribute traffic across multiple regions, not just multiple zones within a region. While this would provide high availability, it would also increase costs.

Option D, performing regular backups of the application and restoring from backups when notified, does not prevent downtime or minimize costs. It only provides a way to recover from downtime after it has occurred.

Therefore, Option A is the best option to adjust the design and support the failure of a single Compute Engine zone, eliminate downtime, and minimize cost.