Google PCA Manage and provision cloud infrastructure Practice Question

Which THREE are best practices for designing a highly available application on Compute Engine?

⚠ Common exam trap

Google Cloud often tests the misconception that local SSDs are suitable for stateful data in HA designs, but the trap is that local SSDs are ephemeral and data is lost on instance failure, so they should only be used for cache or temporary data, not for persistent state.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Use managed instance groups with autoscaling

Option C is correct because a managed instance group (MIG) with autoscaling automatically maintains the desired number of healthy VM instances and replaces failed ones, which is fundamental to high availability on Compute Engine. Option D is correct because an external load balancer with health checks only routes traffic to healthy backends and removes unhealthy instances from rotation, preventing users from hitting failed VMs. Option E is correct because distributing instances across multiple zones protects the application from a single-zone failure, since zonal outages do not affect instances in other zones within the same region. Option A is not appropriate because local SSDs are ephemeral and tied to a single VM, so they cannot store durable stateful data for a highly available design. Option B is not appropriate because a single large machine type creates a single point of failure and cannot provide redundancy or fault tolerance.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
Use local SSDs for stateful data
Why it's wrong here
Local SSDs are ephemeral and tied to a single VM's host, so data is lost if that instance stops or its host fails, breaking availability across zones. They suit scratch space, caches or replicated databases where the application already handles redundancy elsewhere.
✗
Use a single large machine type
Why it's wrong here
A single large machine type concentrates all workloads on one VM, creating a single point of failure with no failover path if that instance or its zone goes down. Large machine types suit vertically scaled, non-redundant workloads such as in-memory analytics.
✓
Use managed instance groups with autoscaling
Why this is correct
Managed instance groups with autoscaling maintain instances across multiple zones, automatically replacing failed VMs and scaling capacity to match demand. This directly satisfies the high-availability requirement by eliminating single points of failure and absorbing load spikes without manual intervention, ensuring continuous service during zone outages or traffic surges.
✓
Use an external load balancer with health checks
Why this is correct
An external load balancer with health checks distributes traffic only to healthy backend instances, automatically routing around failed ones. This removes the load balancer and any single instance as a point of failure, directly satisfying the high-availability requirement.
✓
Distribute instances across multiple zones
Why this is correct
Spreading instances across multiple zones isolates the application from a single-zone failure, since each zone has independent power, cooling and networking. This satisfies the high-availability requirement by ensuring that losing one zone leaves capacity running elsewhere, and it also lets a managed instance group automatically recreate failed instances.

Go deeper

Google PCA Manage and provision cloud infrastructure Practice Question

Why each option matters

Option-by-option breakdown

Related to this question