Cloud Migration Guide: Step-by-Step for 2026

Photo by Tima Miroshnichenko on Pexels
The first wave of enterprise cloud migrations was about getting out of the data center. The 2026 wave is about doing it without bankrupting the IT budget. Migration projects we audited last year ranged from $5,000 (a five-VM SaaS startup) to $5M+ (a 4,000-VM banking estate), and the difference between a successful and a stalled program was almost always discipline — not technology choice.
This guide is the step-by-step playbook we use with mid-market and enterprise clients. We cover the 7 Rs framework, realistic timelines, where the budget actually goes, and the operating model changes that decide whether your cloud bill becomes a savings story or a cautionary tale.
How This Guide Works
The framework here is provider-agnostic but mirrors AWS Migration Acceleration Program, Azure Migrate, and Google Cloud Adoption Framework conventions. We assume you already have an executive mandate and at least a target hyperscaler in mind. The phases — Assess, Mobilize, Migrate, Modernize, Operate — are sequential at the program level but parallel at the workload level. Most enterprises run 6–10 workload waves over 18–24 months.
| Phase | Duration | Key Deliverable | Typical Cost |
|---|---|---|---|
| Assess | 4–8 weeks | Discovery + business case | $30K–$150K |
| Mobilize | 6–12 weeks | Landing zone + ops model | $80K–$400K |
| Migrate | 6–18 months | Workloads moved (waves) | $1K–$10K per VM |
| Modernize | Ongoing | Refactored workloads | Variable |
| Operate | Continuous | FinOps + SRE | 8–15% of cloud spend |
Step 1: Assess — Build a Real Business Case
The single biggest cause of failed migrations is a business case built on list pricing instead of TCO. Discovery should produce four artifacts: an inventory (CMDB-quality), a dependency map, a workload-by-workload TCO comparison, and a risk register. Tools like AWS Application Discovery Service, Azure Migrate, Movere, Cloudamize, or even open-source NetBox plus a Jamf agent can build the inventory in 4–8 weeks.
Plan for honest 3-year TCO that includes professional services, parallel run, training, and FinOps tooling — not just compute and storage. We typically see migrations underwater for the first 12–18 months and net-positive by month 24–30.
Step 2: Pick Your R — The 7 Rs Framework
Each application must be classified as one of:
- Retire — switch off; 10–15% of estates have apps no one uses.
- Retain — keep on-prem (mainframes, regulated workloads).
- Rehost — lift and shift; cheapest, fastest, no cloud-native benefits.
- Replatform — minor optimizations (managed DB, autoscaling).
- Repurchase — replace with SaaS (e.g., on-prem CRM to Salesforce).
- Refactor — re-architect for cloud-native (microservices, serverless).
- Relocate — VMware-to-VMware Cloud or hyperscaler partition.
A healthy portfolio mix is roughly 5% Retire, 10% Retain, 35% Rehost, 25% Replatform, 15% Repurchase, 10% Refactor.
Step 3: Build a Landing Zone
The landing zone is the foundation: multi-account/subscription/project structure, IAM and SSO, networking, logging, and guardrails. AWS Control Tower, Azure Landing Zones, and GCP Cloud Foundation Toolkit all accelerate this. Skip it and you’ll be retrofitting account separation and security baselines under deadline pressure six months in.
A production-grade landing zone typically costs $80K–$400K to build, takes 6–12 weeks, and is the highest-leverage spend in the program.
Step 4: Plan Migration Waves
Group workloads into waves of 5–25 applications based on dependencies. Each wave should follow the same pattern: discover, design, test, cutover, hyper-care, retrospective. Waves should be small enough that one runs every 2–4 weeks once the factory is humming.
Typical wave anatomy: week 1 design, week 2 build, week 3 test, week 4 cutover and hyper-care. The first three waves will be slower; expect 30–40% of the total program duration to elapse before throughput stabilizes.
Step 5: Migrate
Tooling is now mature. AWS Application Migration Service (MGN), Azure Migrate Server Migration, and Google Migrate to Virtual Machines all support agent-based block replication with cutovers measured in minutes. Database migration uses DMS, Azure Database Migration Service, or GCP Database Migration Service for homogeneous moves; logical replication or schema converters for heterogeneous moves (e.g., Oracle to Postgres).
Cutover windows have shrunk dramatically — most workloads can flip with under 30 minutes of downtime. Plan rollback paths anyway; the worst migrations are the ones that succeeded but couldn’t be undone.
Step 6: Modernize After You Land
Don’t refactor during migration. Land it, stabilize it, then modernize. Common second-pass moves include containerizing on EKS/AKS/GKE, swapping self-managed databases for RDS/Cloud SQL/Azure SQL, replacing batch jobs with Step Functions or Cloud Workflows, and pushing static assets to CloudFront or Cloud CDN.
Step 7: Operate — FinOps from Day One
Migrating without a FinOps capability is how teams end up 40% over budget by month 6. Stand up cost allocation tags, shared dashboards, and a weekly FinOps cadence before the first wave moves. Tools like Vantage, CloudHealth, Apptio Cloudability, or native Cost Explorer/Azure Cost Management are non-optional.
Realistic Migration Cost Model
| Estate Size | Timeline | Total Program Cost | Per VM |
|---|---|---|---|
| 50 VMs | 6 months | $250K–$500K | $5K–$10K |
| 250 VMs | 12 months | $750K–$1.8M | $3K–$7K |
| 1,000 VMs | 18 months | $2.5M–$5M | $2.5K–$5K |
| 4,000+ VMs | 24+ months | $7M–$20M | $1.7K–$5K |
| 10,000+ VMs | 30+ months | $15M+ | $1.5K–$4K |
These figures include professional services, tooling, and parallel run — not the post-migration cloud bill itself.
How to Run a Successful Migration
- Get an executive sponsor who can break ties on architecture and budget.
- Run a small, real workload through end-to-end before committing to waves.
- Standardize on three or four reference architectures; refuse snowflakes.
- Move data alongside compute to avoid egress traps.
- Train operators in parallel — your old runbooks won’t work in cloud.
Recommended Offers
💡 Editor’s pick: AWS Migration Acceleration Program (MAP) provides credits and partner-led assessment for qualifying enterprise migrations.
💡 Editor’s pick: Microsoft’s Azure Migrate and Modernize program subsidizes ISV partner-led discovery for estates of 1,000+ servers.
💡 Editor’s pick: Google’s Rapid Migration Program (RaMP) bundles GCP credits with Anthos and Migrate to Containers for hybrid cutovers.
FAQ — Cloud Migration
Q: How long does a typical migration take? A: 6–24 months depending on estate size. Most 250–1,000 VM moves complete in 12–18 months end-to-end.
Q: Should I lift-and-shift or refactor? A: Lift-and-shift first to exit the data center, then modernize selectively. Pure refactor programs run 2–3x longer with no faster value capture.
Q: How much can I save by moving to cloud? A: Realistically, 15–30% of TCO over 3 years if you also right-size, modernize, and run FinOps. Lift-and-shift alone often costs more than on-prem in year one.
Q: What’s the most common migration mistake? A: Underestimating the operating model change. Cloud demands different skills, different runbooks, and different financial governance.
Q: Do I need a partner? A: For estates over 250 VMs or in regulated industries, almost always yes. Partner fees usually run 25–40% of program cost.
Q: How do I avoid the egress trap? A: Move databases and data lakes alongside the apps that read them. Stranded data on-prem with apps in cloud is the most expensive pattern.
Related Reading on ERP Softnic
- Best Cloud Computing Providers of 2026: Top 10 Compared
- AWS vs Azure vs GCP: 2026 Complete Comparison
- Cloud Cost Optimization Strategies for 2026
- Cloud Security Best Practices for 2026
- Multi-Cloud Strategy Guide 2026
Final Verdict
Cloud migration in 2026 is a solved engineering problem and an unsolved organizational one. The technology — replication, automation, landing zones — is mature enough that most failures trace back to weak business cases, missing FinOps, or untrained operators. Run it as a portfolio program with executive sponsorship, build the landing zone first, and modernize after you land. Get those three right and you’ll join the cohort that captures real value, not just a new bill from a different vendor.
This article is for informational purposes only. Cloud pricing, services, and SLAs are accurate as of publication and subject to change. ERP Softnic may receive compensation for some placements; rankings are independent.
By ERP Softnic Editorial · Updated May 9, 2026
- cloud computing
- cloud migration
- 2026
- infrastructure