Migration Strategies

Migrating workloads to Google Cloud is a common scenario that intersects with every pillar of the Architecture Framework. A well-planned migration reduces risk, minimises downtime, and positions workloads for ongoing optimisation on GCP. This lesson covers the migration strategies, tooling, and best practices recommended by the GCP Architecture Framework.

---

The 6 Rs of Migration

Google Cloud and the broader industry recognise six primary migration strategies, often called the "6 Rs":

Strategy	Description	Effort	Cloud Benefit
Rehost (Lift and Shift)	Move workloads to GCP with minimal changes	Low	Quick migration, reduced data centre costs
Replatform (Lift and Optimise)	Make minor adjustments to leverage managed services	Medium	Reduced operational overhead
Refactor (Re-architect)	Redesign applications to be cloud-native	High	Maximum cloud benefits, scalability, resilience
Repurchase	Replace with a SaaS or commercial solution	Variable	Reduced maintenance, vendor-managed updates
Retain	Keep the workload on-premises (not ready to migrate)	None	Avoid unnecessary migration risk
Retire	Decommission the workload entirely	None	Cost savings from eliminating unused systems

---

Choosing a Strategy

Decision Matrix

Factor	Rehost	Replatform	Refactor
Time to migrate	Weeks	Months	Months-Years
Risk	Low	Medium	High
Cost savings	Moderate (hosting only)	Good (hosting + operations)	Best (hosting + operations + efficiency)
Team skills required	Basic cloud knowledge	Cloud services familiarity	Cloud-native development expertise
Performance improvement	Minimal	Moderate	Significant
Ongoing operational burden	High (you still manage VMs)	Medium	Low (managed services handle most)

When to Use Each Strategy

Strategy	Ideal Scenarios
Rehost	Data centre closure deadline, legacy systems with limited documentation, quick wins
Replatform	Database migrations (Cloud SQL), moving to managed Kubernetes (GKE), containerising existing apps
Refactor	Monolith to microservices, adopting serverless (Cloud Run), high-growth applications needing elasticity

---

Migration Phases

Google recommends a phased approach to migration:

Phase 1: Assess

Activity	Description
Inventory	Catalogue all workloads, dependencies, and infrastructure
Assess readiness	Evaluate each workload's migration complexity and cloud suitability
TCO analysis	Compare the total cost of ownership on-premises vs GCP
Risk assessment	Identify migration risks and mitigation strategies
Prioritise	Order workloads by business value, risk, and complexity

GCP Migration Assessment Tools

# Use the Migration Center (formerly StratoZone) for discovery and assessment
# Install the discovery agent on your on-premises servers
# Or use agentless discovery via API scanning

Phase 2: Plan

Activity	Description
Architecture design	Design the target GCP architecture for each workload
Landing zone	Set up the GCP foundation (projects, VPCs, IAM, logging)
Network connectivity	Establish connectivity between on-premises and GCP (Cloud VPN or Interconnect)
Migration runbook	Document step-by-step procedures for each workload migration
Rollback plan	Define how to revert if the migration encounters issues

Phase 3: Deploy (Migrate)

Activity	Description
Pilot migration	Migrate a non-critical workload first to validate the process
Data migration	Transfer data using appropriate methods (online, offline, streaming)
Application migration	Move compute workloads using the chosen strategy
Testing	Validate functionality, performance, and security in the target environment
Cutover	Switch production traffic to the GCP environment

Phase 4: Optimise