Question

A retail company runs a stateless web frontend on several individually created Compute Engine VMs behind a network load balancer. Operators must manually recreate any VM that crashes, and scaling during promotions is done by hand. The architect wants Google Cloud to automatically replace unhealthy instances and add or remove capacity based on CPU utilization, while keeping all instances identical. Which configuration should the architect implement?

Accepted Answer

Create a regional managed instance group from an instance template, enable autohealing with an application health check, and configure autoscaling based on CPU utilization.

Answer

Add the existing VMs to an unmanaged instance group and attach an HTTP load balancer with a health check to route traffic away from failed instances.

Answer

Enable Compute Engine live migration on each VM and set up a Cloud Monitoring alert that emails operators when CPU utilization exceeds a threshold.

Answer

Configure a sole-tenant node group with automatic restart enabled and manually resize the group before each promotional event.