Skip to main content

GCP Patterns

Status

Master Plan - Comprehensive init.md complete, ready for SKILL.md implementation

Google Cloud Platform architectural patterns covering data analytics, machine learning, Kubernetes, and serverless computing.

Scope​

This skill provides GCP-specific service selection and patterns:

  • Compute: Cloud Run vs. GKE vs. Cloud Functions vs. Compute Engine
  • Data & Analytics: BigQuery, Pub/Sub, Dataflow, Dataproc
  • AI/ML: Vertex AI, AutoML, pre-trained APIs, TPUs
  • Storage: Cloud Storage tiers, Persistent Disk, Filestore
  • Databases: Cloud SQL, Cloud Spanner, Firestore, Bigtable
  • Networking: VPC design, Cloud Load Balancing, Cloud CDN, Cloud Armor

Key Components​

  • Cloud Run First - Default choice for stateless HTTP services (serverless, auto-scale to zero)
  • BigQuery for Analytics - Best-in-class petabyte-scale data warehouse
  • GKE for Kubernetes - Most mature managed Kubernetes (invented by Kubernetes creators)
  • Vertex AI Platform - Unified ML lifecycle (training, deployment, monitoring)
  • Multi-Region Architecture - 99.95% SLA with multi-region deployment
  • Workload Identity - Modern authentication for GKE (no service account keys)

Decision Framework​

Compute Service Selection:

HTTP service + stateless → Cloud Run
Need Kubernetes control → GKE
Simple event-driven → Cloud Functions
Full VM control → Compute Engine

Database Selection:

Global distribution → Cloud Spanner
Relational (regional) → Cloud SQL
Document + real-time → Firestore
Key-value + scale → Bigtable

Tool Recommendations​

  • Terraform (hashicorp/google provider) - Multi-cloud IaC
  • gcloud CLI - Official Google Cloud command-line tool
  • google-cloud-* (Python SDK) - Comprehensive API coverage
  • Cloud Code (VS Code/IntelliJ) - IDE integration for GCP

Integration Points​

  • writing-infrastructure-code - GCP resources via Terraform
  • operating-kubernetes - GKE management and Workload Identity
  • architecting-data - BigQuery as data warehouse
  • building-ci-pipelines - Cloud Build for CI/CD
  • secret-management - Secret Manager integration
  • implementing-observability - Cloud Monitoring/Logging (formerly Stackdriver)
  • ai-data-engineering - Vertex AI + BigQuery ML pipelines

Learn More​