Google Cloud Compute Engine Cheatsheet: Cloud Computing Platform

Overview

Google Cloud Compute Engine provides scalable, on-demand virtual machine instances running on Google's global infrastructure, offering flexible, high-performance compute resources with deep integration into the broader Google Cloud ecosystem.

Core Concepts

Infrastructure as a Service (IaaS)
Scalable Compute Resources
Global, Regional, and Zonal Resources
Sustained-use and Committed-use Pricing Models

Key Components

Virtual Machine Instances
- Compute resources with customizable machine types
- Predefined and custom machine configurations
Resource Manager and Deployment Manager
- Project, folder, and organization hierarchy
- Infrastructure as Code via Terraform or Config Connector
Virtual Private Cloud (VPC)
- Global VPC networks
- Subnet configuration per region
- Firewall rules and Cloud Armor

Machine Families and Types

General Purpose (E2, N2, N2D, N4, C4, T2D, T2A)
- Balanced price/performance
- Web servers, small-to-medium databases, dev/test
Compute Optimized (C3, C3D, C4, H3)
- High per-core performance
- HPC, gaming servers, ad serving
Memory Optimized (M1, M2, M3, X4)
- Large in-memory workloads
- SAP HANA, in-memory databases, analytics
Storage Optimized (Z3)
- High local SSD throughput
- Scale-out databases, analytics workloads
Accelerator Optimized (A2, A3, G2)
- NVIDIA GPUs (H100, A100, L4) and TPUs via separate services
- Machine learning, training, inference, HPC

Provisioning Workflow

Bash

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
# gcloud CLI VM creation
gcloud compute instances create my-vm \
    --zone=us-central1-a \
    --machine-type=e2-medium \
    --image-family=debian-12 \
    --image-project=debian-cloud

# Terraform example
resource "google_compute_instance" "my_vm" {
  name         = "my-vm"
  machine_type = "e2-medium"
  zone         = "us-central1-a"
  boot_disk {
    initialize_params { image = "debian-cloud/debian-12" }
  }
  network_interface {
    network = "default"
    access_config {}
  }
}

Key Management Commands

Bash

1
2
3
4
5
6
7
8
9
10
11
# Instance management
gcloud compute instances list
gcloud compute instances start my-vm --zone=us-central1-a
gcloud compute instances stop my-vm --zone=us-central1-a
gcloud compute instances delete my-vm --zone=us-central1-a

# Network configuration
gcloud compute networks create my-vpc --subnet-mode=custom
gcloud compute firewall-rules create allow-ssh \
    --network=my-vpc --allow=tcp:22

Networking Configurations

Global Virtual Private Cloud (VPC) networks
Regional subnets with secondary IP ranges
VPC firewall rules and hierarchical firewall policies
Cloud Load Balancing (Global, Regional, Internal)
Cloud VPN, Cloud Interconnect, Cross-Cloud Interconnect
External, internal, and IPv6 addresses
Private Service Connect

Storage Options

Persistent Disk and Hyperdisk
- Network-attached block storage
- Balanced, SSD, Extreme, and Hyperdisk (Balanced, Throughput, Extreme, ML)
- Snapshots and asynchronous replication
Local SSD
- High-IOPS, low-latency local NVMe storage
- Ephemeral; data lost on stop or migration
Cloud Storage / Filestore
- Object and managed NFS storage for shared workloads

Security Best Practices

Use Cloud IAM with least privilege
Apply organization policies and IAM Conditions
Enable Confidential VMs and disk encryption (CMEK/CSEK)
Use OS Login with 2-Step Verification instead of project SSH keys
Configure VPC firewall rules and Cloud Armor
Monitor with Security Command Center and Cloud Audit Logs

Cost Optimization Strategies

Committed Use Discounts (resource-based and flexible)
Spot VMs (successor to Preemptible VMs)
Sustained Use Discounts (automatic)
Autoscaling and right-sizing recommendations
Custom machine types to match workload
Active Assist and Recommender insights

Performance Optimization

Choose the appropriate machine family (e.g., C3 for compute, M3 for memory)
Use Hyperdisk or SSD persistent disks
Tune disk performance independently of VM size (Hyperdisk)
Apply global or regional load balancing
Monitor with Cloud Monitoring and Ops Agent

Scaling Techniques

Horizontal Scaling
- Managed Instance Groups (MIGs)
- Regional MIGs with autohealing and autoscaling
Vertical Scaling
- Change machine type while stopped
- Move between predefined and custom types