Evaluating cloud infrastructure? Learn about our Software Development services.
Read also: Technical Due Diligence Checklist for Software Projects | Cloud Migration Checklist: Step-by-Step Planning Guide
The question is no longer whether to move to the cloud — it is whether the business case justifies it for your specific workloads. Too many migration decisions are made on vendor marketing rather than hard numbers. This guide provides a structured framework for calculating the true cost of migrating from on-premise infrastructure to AWS or Azure, including the hidden expenses that turn a projected 30% savings into a 15% cost increase.
The TCO Comparison Framework
A valid Total Cost of Ownership comparison must include six cost categories. Most organizations calculate only the first two and wonder why their cloud bill exceeds projections.
Category 1: Infrastructure costs
On-premise:
| Cost item | Typical range (per server/year) | Notes |
|---|---|---|
| Server hardware (amortized over 5 years) | $2,000-$6,000 | Includes refresh cycle |
| Storage (SAN/NAS, amortized) | $1,000-$4,000 | Per TB, depends on tier |
| Networking equipment | $500-$2,000 | Switches, firewalls, load balancers |
| Data center space | $1,000-$5,000 | Power, cooling, physical security |
| Backup infrastructure | $500-$2,000 | Tape, disk-to-disk, offsite |
Cloud (AWS/Azure equivalent):
| Cost item | Typical range (per instance/year) | Notes |
|---|---|---|
| Compute (reserved instances, 1-year) | $2,500-$15,000 | Varies by instance type |
| Storage (EBS/Managed Disks) | $1,200-$6,000 | Per TB, varies by tier |
| Networking (VPC, load balancers) | $600-$3,000 | Data transfer extra |
| Backup (snapshots, cross-region) | $300-$1,500 | Incremental snapshots |
| Data egress | $500-$5,000 | Often the surprise line item |
Category 2: Operations and personnel
On-premise infrastructure requires a team. Cloud computing shifts responsibilities but does not eliminate them.
On-premise team (typical for 50-100 servers):
- 2-3 system administrators: $180,000-$360,000/year
- 1 network engineer: $90,000-$140,000/year
- 1 storage specialist (part-time or shared): $40,000-$70,000/year
- 1 security engineer (shared): $50,000-$80,000/year
Cloud team (equivalent workload):
- 1-2 cloud engineers / DevOps specialists: $120,000-$280,000/year
- 1 cloud security engineer (shared): $50,000-$80,000/year
- Cloud architect (fractional or consulting): $30,000-$80,000/year
Cloud typically reduces headcount by 30-50% for infrastructure operations, but the remaining roles require higher-paid specialists with cloud-specific expertise.
Category 3: Software licensing
This category causes the most budget surprises. Many enterprise software licenses — particularly Microsoft SQL Server, Oracle Database, and SAP — have different pricing models for cloud environments.
Common licensing traps:
- Oracle Database: On-premise licensing is per-processor. On AWS/Azure, Oracle counts vCPUs differently, often doubling or tripling the license cost
- Microsoft SQL Server: Bring-your-own-license (BYOL) requires Software Assurance. Without it, you pay cloud-native pricing at a premium
- Windows Server: License mobility rules changed in 2022. Verify eligibility before assuming your licenses transfer
- Third-party tools: Monitoring, backup, and security tools often have separate cloud pricing tiers
Category 4: Migration execution costs
The migration itself is a project with its own budget. These are one-time costs, but they are significant.
| Migration activity | Cost range | Notes |
|---|---|---|
| Assessment and planning | $15,000-$50,000 | Application discovery, dependency mapping |
| Architecture design | $10,000-$40,000 | Target architecture, security model |
| Application refactoring | $20,000-$200,000 | Depends on application count and complexity |
| Data migration | $10,000-$50,000 | Data transfer, validation, cutover |
| Testing and validation | $15,000-$60,000 | Functional, performance, security testing |
| Training | $5,000-$25,000 | Cloud operations, new tools |
| Parallel running | $20,000-$80,000 | Dual infrastructure during transition |
Category 5: Ongoing optimization costs
Cloud cost optimization is not a one-time activity. Without continuous management, cloud spending drifts upward by 15-30% annually.
- FinOps practice: $50,000-$150,000/year for tooling and dedicated personnel
- Reserved instance management: Requires quarterly review and commitment decisions
- Right-sizing: Monthly review of instance utilization and adjustment
- Waste elimination: Identifying and removing unused resources, orphaned storage, idle load balancers
Category 6: Risk and opportunity costs
Cloud advantages:
- Faster time-to-market for new applications (weeks vs months for infrastructure provisioning)
- Elastic scaling for variable workloads (pay only for what you use)
- Disaster recovery at a fraction of on-premise cost
- Access to managed services (AI/ML, analytics, IoT) without building from scratch
Cloud risks:
- Vendor lock-in — switching cloud providers is expensive and disruptive
- Data sovereignty — some regulations require data residency in specific jurisdictions
- Outage exposure — single-cloud dependency creates availability risk
- Cost unpredictability — without governance, cloud bills can spike unexpectedly
The 5-R Migration Strategy and Cost Impact
Not every workload should be migrated the same way. The migration strategy directly determines cost.
| Strategy | Description | Relative cost | Best for |
|---|---|---|---|
| Rehost (lift-and-shift) | Move as-is to cloud VMs | Low ($2K-$10K/app) | Legacy apps with minimal cloud benefit |
| Replatform (lift-and-reshape) | Minor adjustments for cloud services | Medium ($10K-$30K/app) | Databases, middleware |
| Refactor (re-architect) | Rebuild using cloud-native services | High ($30K-$150K/app) | Core business apps with scaling needs |
| Repurchase | Replace with SaaS equivalent | Variable | CRM, ERP, HR systems |
| Retire | Decommission | Savings | Unused or redundant apps |
Most enterprise migrations use a mix of all five strategies. A typical distribution: 40% rehost, 25% replatform, 15% refactor, 10% repurchase, 10% retire.
Building Your Cost Model: Step by Step
Step 1: Inventory current infrastructure
Document every server, application, database, and network device. For each item capture: CPU cores, RAM, storage (type and capacity), network throughput, average and peak utilization, and operating system/software licenses.
Step 2: Map applications to migration strategies
Classify each application using the 5-R model. This is where most organizations need external expertise — the classification requires understanding both the application architecture and cloud service options.
Step 3: Calculate on-premise TCO (3-year)
Sum all six cost categories for your current environment, projected over 3 years. Include planned hardware refreshes, salary increases, and data center contract renewals.
Step 4: Calculate cloud TCO (3-year)
Use the AWS Pricing Calculator or Azure TCO Calculator as a starting point, then add the costs those tools exclude: migration execution, training, parallel running, licensing changes, and ongoing optimization.
Step 5: Apply realistic adjustment factors
- Add 20-30% contingency to the migration execution budget
- Add 15-25% to Year 2 and Year 3 cloud costs for organic growth
- Subtract 10-20% from cloud costs if you commit to reserved instances and active FinOps
Step 6: Model the transition period
During migration, you pay for both environments. Model a 6-12 month overlap period with gradual reduction in on-premise costs as workloads move.
Optimization Strategies That Actually Work
Reserved instances and savings plans
Committing to 1-year or 3-year reserved instances reduces compute costs by 30-60% compared to on-demand pricing. Start with 1-year commitments for production workloads with stable usage patterns.
Right-sizing
Most organizations over-provision cloud instances by 40-60%. Right-sizing — matching instance type and size to actual workload requirements — typically reduces compute costs by 25-35%.
Spot instances for non-critical workloads
Development, testing, and batch processing workloads can use spot instances at 60-90% discount. Design workloads for interruption tolerance and the savings are substantial.
Auto-scaling
Configure auto-scaling for variable workloads instead of provisioning for peak demand. Web applications, API backends, and event-driven workloads benefit most.
Storage tiering
Move infrequently accessed data to lower-cost storage tiers. AWS S3 Intelligent-Tiering and Azure Blob Storage tiering can reduce storage costs by 40-70% for data with variable access patterns.
How ARDURA Consulting Accelerates Cloud Migration
Cloud migration requires a team with deep expertise across infrastructure, DevOps, security, and application architecture — skills that are scarce and expensive to hire permanently.
- 500+ senior specialists across cloud architecture, DevOps, security engineering, and backend development — available within 2 weeks
- 40% cost savings compared to traditional hiring, with the flexibility to scale your migration team as needed
- 99% client retention — your cloud team stays consistent through assessment, migration, and optimization phases
- 211+ completed projects — engineers who have migrated enterprise workloads before and know where the cost traps are
From a single cloud architect to assess your migration readiness to a full migration squad, ARDURA Consulting provides the expertise that turns cost projections into realized savings.