Cloud Modernization Services

Bill Shock from Lift-and-Shift

Moving VMs without re-architecting creates $2–5x cost increases — cloud bills immediately exceed on-prem costs. One documented pattern: a 200-server lift-and-shift produced an $800K/year AWS bill versus $300K/year on-prem. The workloads were identical; the billing model was not.

Prevention: Require a FinOps plan and cloud cost model before migration begins, not after. The cost model must include reserved instance requirements, storage tier decisions, and data transfer costs — not just compute comparisons.

Egress Trap

Vendors and cloud hyperscalers don't emphasise egress pricing in sales conversations. At $0.08–0.09/GB for outbound data, a 100TB/month workload generates $8,000–$9,000/month in egress fees alone — $100K/year that was absent from the TCO model. This is not an edge case; it's the median outcome for data-intensive workloads.

Prevention: Include egress cost modelling in every TCO analysis. Consider data transfer patterns before selecting the primary cloud provider — workloads with heavy outbound traffic to on-prem or third parties have fundamentally different cloud economics.

Repatriation Spiral

18 months after migration, unexpected costs and complexity drive repatriation. A 2024 study found 53% of enterprises moved at least one workload back on-prem in the prior two years. The primary drivers are egress costs, Microsoft SQL Server licensing premiums (3–5× more expensive on cloud), and compliance requirements discovered post-migration.

Prevention: Realistic TCO analysis including egress, reserved instance commitment requirements, support costs, and software licensing premiums on cloud — not just compute comparisons. Repatriation is an avoidable cost if the TCO model is honest.

No FinOps plan proposed

cloud cost optimisation treated as an afterthought rather than a first-class architectural concern. If the proposal doesn't include a cost model with reserved instance strategy and egress analysis, the first cloud bill will be a surprise.

"Cloud-native" claim without evidence

"cloud-native" in a proposal without specific serverless, container, or managed service evidence in comparable case studies is marketing language. Ask for a workload that went through re-architecture, not just migration, with before/after cost data.

Fixed-price contract for open-ended refactoring scope

refactoring complexity is discovered during migration, not before. Fixed-price contracts for re-architecture work create incentives to cut scope, not deliver value. Time-and-materials with milestone-based caps is the appropriate commercial model.

Cloud-agnostic claim without multi-cloud delivery evidence

most implementation firms have a primary hyperscaler relationship that drives their recommendations. Ask directly which hyperscaler generated their last five client migrations and what percentage of revenue comes from each cloud's partner programme.

Migration plan without parallel-run or rollback strategy

critical workloads migrated without a parallel-run period or tested rollback procedure create outage risk with no recovery path. Any proposal that skips rollback planning for Tier 1 systems is accepting production risk on your behalf.

How much does cloud migration cost?

Migration projects run $150K–$2M+ depending on workload count and re-architecture scope. Lift-and-shift projects for 50–200 workloads typically run $200K–$600K. Re-architecture or cloud-native refactoring is $500K–$2M+. Budget 20–30% above estimates for undiscovered dependencies and parallel-run costs.

AWS vs Azure vs GCP — which should we choose?

Workload profile determines this more than vendor preference. AWS leads for greenfield and startup workloads; Azure dominates where Microsoft licensing (Office 365, Active Directory, Dynamics) is already heavy; GCP is strongest for data/AI workloads and Kubernetes-native architectures. Multi-cloud adds 30–40% operational overhead — most organisations benefit from a primary cloud strategy.

How long does a cloud migration take?

A 50-server lift-and-shift takes 3–4 months. A 200-server re-architecture takes 12–18 months. Full data centre exit for 500+ workloads takes 18–36 months. Projects that try to compress timelines by skipping the assessment phase typically add 4–8 months through rework.

What is FinOps and do we need it?

FinOps is the practice of cloud cost optimisation — matching cloud spend to business value. Without it, cloud bills average 35% higher than necessary. You need it before migration (for TCO modelling), during migration (to avoid over-provisioning), and ongoing (for right-sizing and reserved instance management). A FinOps platform (CloudHealth, Apptio, native tooling) costs $50K–$150K/year but typically saves 20–35% of cloud spend.

Should we re-architect or lift-and-shift?

Lift-and-shift is faster (3–6 months) but creates an ongoing cost premium and misses cloud-native benefits. Re-architecture is 6–18 months but delivers elastic scaling, managed service cost savings, and developer velocity improvements. Most enterprises start with lift-and-shift for non-critical systems to build cloud muscle, then re-architect strategic workloads.

What is the repatriation risk — could we need to move back on-prem?

Repatriation is real: 53% of enterprises moved at least one workload back in the past two years (2024 data). The primary drivers are unexpected egress costs, licensing complexity (Microsoft SQL Server on cloud is 3–5× more expensive), and compliance requirements that emerged post-migration. A proper TCO model before migration prevents 80% of repatriation decisions.