Google Cloud ACE Compute Engine and MIGs Guide

Study Google Cloud ACE Compute Engine and MIGs: key concepts, common traps, and exam decision cues.

This lesson is about the practical VM operations Google Cloud expects ACE candidates to recognize quickly. The exam often asks whether the real need is a one-off instance, a managed instance group, OS Login control, or fleet management through VM Manager.

Managed instance group (MIG): Group of similar VM instances managed together for scaling, health, or consistent rollout.

OS Login: Google Cloud feature that uses IAM to control SSH access to Compute Engine instances.

VM Manager: Google Cloud management layer for patching, inventory, configuration, and fleet-level VM administration.

What Google Cloud is really testing here

ACE wants you to separate:

  • one-off VM deployment from fleet deployment
  • instance template reuse from manual instance-by-instance setup
  • IAM-governed SSH access from unmanaged key sprawl
  • VM fleet administration from simple instance creation

The exam often disguises this as a generic “deploy a VM” question, but the real answer is usually about the operating pattern around that VM.

Fast compute chooser

If the question is mainly about… Strongest first lane
one machine with a specific one-time purpose single Compute Engine instance
many similar VMs that should scale or heal together managed instance group
SSH access controlled through IAM instead of metadata-based keys OS Login
patching, inventory, and fleet maintenance across many VMs VM Manager

Single instance versus MIG versus VM Manager

Control What it really answers
Compute Engine instance how to launch one VM
MIG how to run many similar VMs from a shared template
OS Login who is allowed to log in and with what IAM-controlled access
VM Manager how to observe and maintain VMs at fleet scale

Candidates often pick MIG when the requirement is really just one VM, or they pick VM Manager when the stem is really about deployment pattern rather than fleet maintenance.

MIG versus one-off instance

Question Single instance MIG
Main purpose one specific VM repeatable fleet of similar VMs
Strongest first when there is no scaling or fleet pattern in the stem the workload should autoscale, autoheal, or roll out consistently
Common trap trying to hand-manage a fleet one VM at a time choosing MIG when the stem only needs a single bespoke host

Managed instance groups are strong on ACE because Google explicitly treats them as the scalable, repeatable Compute Engine lane. The moment the stem starts talking about multiple identical VMs, shared templates, or fleet healing, MIG should move to the front of your elimination order.

OS Login versus unmanaged SSH keys

ACE does not expect deep Linux internals here. It expects you to notice when the real issue is access governance, not compute provisioning.

Use OS Login when:

  • SSH access should follow IAM permissions
  • local key sprawl should be reduced
  • project-level or instance-level access should be managed centrally

That is stronger than hand-managing metadata-based SSH keys when the stem clearly values access control and auditability.

Common traps

Trap Better reading
“A group of VMs is just several instances.” A managed fleet with scaling or healing needs a MIG, not hand-managed singles.
“OS Login is just another SSH key store.” OS Login ties SSH access to IAM instead of relying on unmanaged metadata-key patterns.
“VM Manager deploys workloads.” VM Manager is about fleet operations and maintenance, not replacing core deployment choices.
“MIG and load balancer mean the same thing.” The MIG is the fleet pattern; the load balancer is the traffic distribution layer.

Harder scenario question

A team needs a regional fleet of identical web VMs that should recreate failed instances automatically and grow with demand. Which lane is strongest first?

  • A. One manually configured VM in each zone
  • B. Managed instance group
  • C. Cloud DNS only
  • D. BigQuery scheduled query

Correct answer: B. The requirement is a shared-template, autohealing, scaling fleet. That is the MIG lane, not a collection of manually maintained instances.

Decision order that usually wins

  1. Separate fleet creation, identity-based instance access, and ongoing VM maintenance.
  2. If many similar VMs should autoscale from a shared definition, think managed instance group.
  3. If VM access should follow IAM identities rather than local keys, think OS Login.
  4. If the issue is patching, inventory, or fleet maintenance, think VM Manager.
  5. ACE usually rewards the tool that matches the VM lifecycle stage directly instead of stacking unrelated features.

Quiz

Loading quiz…
Revised on Sunday, May 10, 2026