GCP Compute Engine Cheat Sheet
Commands for creating, connecting to, and managing virtual machine instances and disks on Google Compute Engine.
2 PagesBeginnerFeb 15, 2026
Instance Lifecycle
Create, list, and control VM instances.
bash
gcloud compute instances create my-vm \ --zone=us-central1-a \ --machine-type=e2-medium \ --image-family=debian-12 --image-project=debian-cloud \ --boot-disk-size=20GBgcloud compute instances listgcloud compute instances stop my-vm --zone=us-central1-agcloud compute instances start my-vm --zone=us-central1-agcloud compute instances delete my-vm --zone=us-central1-a
SSH & File Transfer
Connect to and copy files to/from an instance.
bash
gcloud compute ssh my-vm --zone=us-central1-agcloud compute scp ./localfile.txt my-vm:~/remote.txt --zone=us-central1-agcloud compute scp my-vm:~/remote.txt ./localfile.txt --zone=us-central1-a
Disks & Images
Manage persistent disks and custom images.
bash
gcloud compute disks create my-disk --size=100GB --zone=us-central1-agcloud compute instances attach-disk my-vm --disk=my-disk --zone=us-central1-agcloud compute images create my-image --source-disk=my-disk --source-disk-zone=us-central1-a
Machine Families
Key machine families to know.
- E2- Cost-optimized, general purpose, shared-core options available
- N2 / N2D- Balanced price/performance for general workloads
- C2 / C3- Compute-optimized, high per-core performance
- A2 / A3- GPU-accelerated instances for ML/AI workloads
- Preemptible / Spot VM- Short-lived, discounted instances that can be reclaimed by GCP anytime
Key Concepts
Key key concepts to know.
- Instance Template- Reusable VM configuration used to create instances or managed instance groups
- Managed Instance Group (MIG)- Auto-scaling, self-healing group of identical VM instances
- Persistent Disk- Durable network-attached block storage independent of instance lifecycle
- Startup Script- Shell script run automatically on instance boot via metadata
Instance Template + Managed Instance Group
Define a reusable template and an auto-healing, auto-scaling MIG from it.
bash
gcloud compute instance-templates create web-template \ --machine-type=e2-medium \ --image-family=debian-12 --image-project=debian-cloud \ --metadata-from-file=startup-script=startup.shgcloud compute instance-groups managed create web-mig \ --base-instance-name=web \ --template=web-template \ --size=3 --zone=us-central1-agcloud compute instance-groups managed set-autoscaling web-mig \ --zone=us-central1-a \ --max-num-replicas=10 --min-num-replicas=2 \ --target-cpu-utilization=0.6 --cool-down-period=90
Health Check + Load-Balanced Backend Service
Wire a MIG to an HTTP(S) load balancer with automatic instance replacement on failed checks.
bash
gcloud compute health-checks create http web-hc \ --port=80 --request-path=/healthz \ --check-interval=10s --unhealthy-threshold=3gcloud compute instance-groups managed update web-mig \ --zone=us-central1-a --health-check=web-hc --initial-delay=60gcloud compute backend-services create web-backend \ --global --protocol=HTTP --health-checks=web-hcgcloud compute backend-services add-backend web-backend \ --global --instance-group=web-mig --instance-group-zone=us-central1-a
Per-Instance Metadata & Guest Attributes
Pass config into a VM via metadata and read it back from inside the guest.
bash
# Set custom metadata at creationgcloud compute instances create my-vm \ --zone=us-central1-a \ --metadata=app-env=production,release=v2.3.1# Inside the VM, read metadata via the metadata servercurl -H "Metadata-Flavor: Google" \ "http://metadata.google.internal/computeMetadata/v1/instance/attributes/app-env"# Query it back externally without SSHgcloud compute instances describe my-vm --zone=us-central1-a \ --format='get(metadata.items[0].value)'
Shielded VM & Confidential Computing Flags
Harden a VM against boot-level rootkits and encrypt memory in use.
bash
gcloud compute instances create secure-vm \ --zone=us-central1-a \ --machine-type=n2d-standard-4 \ --shielded-secure-boot \ --shielded-vtpm \ --shielded-integrity-monitoring \ --confidential-compute-type=SEV \ --maintenance-policy=TERMINATE
Networking & Placement Concepts
Concepts that matter once you move past a single default-network VM.
- VPC Network Tags- Labels attached to instances used as firewall-rule targets instead of per-IP rules
- Alias IP Ranges- Additional IP ranges routable to a VM's NIC, commonly used for container IPs (e.g. GKE)
- Sole-Tenant Nodes- Dedicated physical servers for instances with licensing or compliance isolation requirements
- Placement Policy- Controls physical proximity (compact) or spread (max distribution) of instances for latency or fault tolerance
- Live Migration- GCP transparently moves a running VM off hardware needing maintenance without rebooting it (standard VMs only, not Spot)
- OS Login- IAM-managed SSH key/user access to instances instead of manually managed project/instance SSH metadata keys
Pro Tip
Use Spot VMs for fault-tolerant batch or stateless workloads — they cost up to 60-91% less than standard on-demand pricing, but design your workload to checkpoint progress since GCP can reclaim them with only 30 seconds notice.
Was this cheat sheet helpful?
Explore Topics
#GCPComputeEngine#GCPComputeEngineCheatSheet#CloudComputing#Beginner#InstanceLifecycle#SSHFileTransfer#DisksImages#MachineFamilies#CheatSheet#SkillVeris
