The five decisions that matter

  • Why are you migrating? Cost, scale, AI, a data center nearing the end of its life, or VMware: the goal determines everything else. → Section 2

  • Which workloads will be moved, and which will remain? → Section 4

  • Which cloud? Hyperscaler, European cloud, private cloud, or a mix. → Section 5

  • Which approach should be taken for each workload? From migrating as-is to rebuilding. → Section 6

  • How do you keep it manageable afterward? With a Landing Zone, FinOps, security, and an exit plan. → Sections 7, 10, and 11

1. What is Cloud Migration?

Cloud Migration means moving applications, data, and infrastructure to a cloud platform. This typically involves moving from a company’s own data center or colocation facility to a public cloud such as AWS, Microsoft Azure, or Google Cloud. It can also involve moving from one cloud to another, or back to an on-premises environment. The latter is called repatriation.

Cloud Migration is different from Application Modernization. With migration, you change where an application runs; with modernization, you change the application itself. Many projects combine the two: first migrate, then modernize each application, or vice versa. You can read about how to decide which approach to take for each application in the Application Modernization guide.

2. Why organizations migrate, and why it doesn’t always go as planned

The most common reasons:

  • the data center or the hardware needs to be replaced;
  • VMware's license terms have changed since its acquisition by Broadcom;
  • The load varies greatly, with peaks that you cannot handle efficiently in your own data center;
  • You want to build AI applications; Gartner expects that by 2029, half of all cloud capacity will be dedicated to AI workloads, compared to less than 10% today;
  • Teams want to deploy faster and use managed services instead of managing everything themselves.

Yet the reality is often disappointing. Gartner expects that by 2028, 25% of organizations will be highly dissatisfied with their cloud adoption, and that by 2029, more than half will fail to achieve the expected results from multicloud. The causes are almost always the same: Moving everything like-for-like and running capacity 24/7, not assigning ownership of the costs to anyone, setting up the infrastructure after the fact, and forgetting about outbound data traffic and licenses. Section 12 shows how to prevent this.

3. Public, private, hybrid, or multicloud

 

Model What It Is Best for
Public cloud A provider's infrastructure and services—you pay based on usage Fluctuating workload, rapid growth, managed services
Private cloud A cloud platform for a single organization, either in its own data center or with a hosting provider Constant load, strict requirements for location or control
Hybrid cloud Public and private clouds integrated, with shared identities, networking, and management Most medium-sized and large organizations
Multicloud Multiple public clouds running side by side Specific services by cloud, distribution of dependencies

In practice, hybrid is the norm: Gartner predicted in 2024 that 90% of organizations would be using a hybrid cloud by 2027. Multicloud requires more expertise and management than it appears. Choose it for a specific reason, such as a service offered exclusively by a single provider. As a general safeguard against lock-in, it is expensive. The difference between hybrid and multicloud is also explained in the frequently asked questions.

4. Which workloads are moving to the cloud?

Start by taking inventory of applications, dependencies, usage, licenses, and owners. Then, for each workload, assess the following:

Characteristic Advocates for the cloud Argues in favor of staying (or later)
Load Variable, with peaks Consistent and predictable
Remaining lifespan Will continue to be used and developed for years to come Will be replaced soon
Licenses Bring your own license, or use a SaaS alternative More expensive in the cloud, or not allowed
Latency and Location Users and integrations are already in the cloud Strict latency requirements for machines, factories, or labs
Data Manageable volume, low outbound traffic Very large datasets with a high volume of outbound traffic
Compliance Requirements met in the selected cloud and region Data must not be transferred outside the organization's control or jurisdiction

Start with workloads that have few dependencies and aren't business-critical. That way, your team can get familiar with the new environment before the heavy-duty systems are migrated.

5. Hyperscaler or European cloud?

That choice has changed over the past two years. Digital sovereignty is now on the agenda of executives and regulators, and Gartner expects European spending on sovereign cloud IaaS to rise by 83% to $12.6 billion in 2026. In July 2026, the Dutch government tightened its cloud policy: anyone using the public cloud must justify the risks and dependence on a single provider and have an exit plan. For critical tasks, the government advises against using providers subject to non-European legislation.

You make that decision on a per-dataset basis. A lot of data can remain with a hyperscaler in a European region, with encryption using your own keys and robust contracts. The most sensitive data requires a European or on-premises environment. A Data Sovereignty Assessment can determine this in just one week.

The AWS European Sovereign Cloud has been available since 15 January 2026, with its first region in Brandenburg. AWS has announced sovereign Local Zones, including one in the Netherlands. It is a separate environment: you migrate to it just as you would to a new region. Check for each service to see if it is already available there.

SUE acts as an Advanced Partner for migrations to AWS; see AWS Consultancy.

Hyperscaler, sovereign, or European cloud

Option Features Watch out
AWS, Azure, or Google Cloud in a European region The widest range of managed services, AI, and tools Is subject to U.S. laws, such as the CLOUD Act
A hyperscaler's sovereign offering, such as the AWS European Sovereign Cloud (since January 2026) Separate infrastructure in the EU, managed by EU residents A smaller range of services; still new. You stay with the same provider: this reduces the jurisdictional risk, not the lock-in
European cloud providers, such as STACKIT, OVHcloud, IONOS, or Scaleway European jurisdiction; in April 2026, the national government entered into a framework agreement with STACKIT Fewer managed services, fewer tools, and less expertise in the market
Private cloud or on-premises Full control over location and access In-house management and investments in hardware

6. Migration paths by workload

Choose a path for each workload. Rehosting (lift-and-shift) is the fastest option, but rarely results in cost savings in the cloud. Replatforming and refactoring make better use of the cloud, but require more work.

All seven routes, along with a table and a decision matrix, are included in the Application Modernization guide.

When migrating, three questions carry particular weight

  • Do you need to act quickly— due to a data center exit or an expiring contract? Then rehost or relocate first, and then modernize.

  • Does the application run 24/7 at a constant load? If so, be sure to carefully calculate the cost of rehosting: without rightsizing and reservations, the cloud is often more expensive.

  • Will the application be phased out within two years? Keep it as is or replace it with SaaS.

7. The Landing Zone

A landing zone is a pre-configured base in the cloud: account structure, identities and roles, network, central logging, security settings, and cost guardrails. Set it up before migrating your first workload. Doing it afterward costs much more, because you’ll have to migrate existing workloads without interrupting them.

  • Define the landing zone as code—for example, using Terraform, Bicep, or CloudFormation—with version control.
  • Use your cloud platform's standard as a starting point: AWS Control Tower or Landing Zone Accelerator, Azure Landing Zones, or the Google Cloud blueprint.
  • Manage tags and budgets from day one so you can see the costs per team or application.
  • Link identities to your existing directory, such as Entra ID, and enforce MFA.

Learn more about the SUE approach on the Landing Zone page.

8. Step-by-step guide

Migrate in waves, not all at once. Each wave teaches your team something for the next one, and having a fallback plan for each wave keeps the risk low.

In eight steps

  • Determine the goal and the constraints: why you are migrating, what the budget is, and what requirements apply.

  • Take inventory of applications, dependencies, usage, and costs. Discovery tools are helpful, but discussions with application owners are essential.

  • For each workload, select a target platform and a migration path, and develop the business case.

  • Set up the landing zone.

  • Migrate a pilot: a single workload with few dependencies.

  • Migrate in waves of similar workloads, with a runbook, a test plan, and a fallback scenario for each wave.

  • Do not shut down the old environment until the new one is running stably.

  • Optimize: rightsizing, reservations or savings plans, autoscaling, and removing what is no longer in use.

9. VMware exit

Since Broadcom’s acquisition of VMware, the licensing terms have changed, and many organizations are looking for an alternative. vSphere 8 and ESXi 8 will be supported through 11 October 2027. Information on what you can do after that is available on the vSphere 8 end of life page. Gartner expects that by 2028, more than one-third of VMware workloads will be running on a different platform, and warns that a complete migration will take three years or longer.

Start with the applications you were already planning to modernize and migrate the rest in phases. At NXP, SUE chose Apache CloudStack, which reduced virtualization costs and provided strategic flexibility. Linux alternatives to vSphere are covered in the guide on choosing a Linux distribution, and virtualization on Kubernetes is covered in the guide on Kubernetes.

Five ways to move away from VMware

Route When Watch out
To the public cloud Workloads that benefit from scale or managed services Calculate the costs of rightsizing and licensing in advance
VMware service from a cloud provider Quickly move out of your own data center without modifying the VMs Time savings; you remain dependent on VMware licenses
Other virtualization platforms, such as OpenShift Virtualization, SUSE Virtualization, Apache CloudStack, Proxmox VE, or KVM with libvirt (also possible: Nutanix, Hyper-V, or Azure Local) Many VMs that remain on-premises New knowledge and tools for your team
To containers Applications You Want to Modernize Anyway More work per application
Staying on VMware Cloud Foundation If the benefits of the switch don't outweigh the costs right now Higher licensing fees; plan your next step

10. How much does a Cloud Migration cost?

 

Cost item What's Inside
Migration Assessment, landing zone, migration work, tests, and a period during which you pay for both the old and the new
Cloud Usage Compute, storage, networking, and outbound data traffic; managed services and support
Licenses Software you transfer or repurchase; some cloud-based licenses are more expensive
Management Platform management, security, monitoring, and FinOps—handled by your own team or a partner
Benefits Eliminated data center and hardware, reduced administrative workload, faster deployment

Plan for three to five years and factor in the costs of doing nothing, such as hardware replacement or a VMware renewal. For larger migrations to AWS, the AWS Migration Acceleration Program may cover a portion of the migration costs. After the migration, FinOps determines whether the business case holds up: at Colect, a three-day analysis resulted in 43% lower monthly AWS costs. Learn more in the guide on reducing cloud costs.

11. Security, compliance, and exit strategy

A migration shifts responsibilities, but it doesn't eliminate them. Therefore, manage security and compliance in the landing zone, and determine how you will exit before the migration.

The Data Act has applied to cloud services since 12 September 2025. You must give notice of a switch with a notice period of no more than two months; the transition takes no longer than thirty days; and, as of 12 January 2027, cloud providers may no longer charge switching fees. The ACM is developing guidelines on which fees will still be permitted.

What to look for

  • Shared responsibility. The provider secures the cloud itself; you secure what you put in it: identities, configurations, data, and applications. Learn more in the guide to cloud security.

  • The Dutch Cybersecurity Act (NIS2). As of 15 August 2026, organizations subject to this law must assess their suppliers and supply chain risks. This includes your cloud provider and migration partner.

  • DORA. As of 17 January 2025, financial institutions must manage their IT risks and third parties, including an exit strategy for critical services.

  • Exit strategy. Before migrating, define how you will exit the cloud. The National Cloud Policy now requires this for key cloud services. The details are listed below.

What does an exit plan include?

An exit plan specifies, for each workload, how you will move to another cloud or back to an on-premises environment: which dependencies you will phase out and how you will transfer your data. Draw it up before the migration. After that, lock-in is already embedded in your architecture and your contracts.

  1. Dependencies per workload. Managed databases, serverless functions, identity in IAM or Entra ID, add-ons in your Kubernetes cluster, and the contract. That is where the lock-in sits, rarely in the virtual machines.
  2. What you consciously accept. A managed database that saves your team a lot of administrative work might be worth it. Write down why, and what leaving would cost.
  3. Data export. Format, volume, and the time required for the export. For large volumes, bandwidth determines how long the migration takes, not the notice period.
  4. The environment in code. Infrastructure as Code (Terraform or OpenTofu) and containerized workloads, so you can rebuild the environment elsewhere.
  5. Terms and fees. Notice period, transition period, and what fees the provider is still allowed to charge. The Data Act sets limits on these (see above).

Test the plan. At least once a year, restore a workload along with its data in a different environment, and record how long it took. Under DORA, financial institutions are already required to have an exit strategy for critical IT services.

SUE incorporates the exit plan into the migration roadmap of the Cloud Readiness Assessment. The free Data Sovereignty Assessment identifies, within one week, which data can be stored where. The page sovereign cloud without vendor lock-in explains where lock-in occurs at each layer in AWS, Azure, and Google Cloud, and outlines the four available paths.

12. Why migrations fail

 

Cause How to Prevent It
Unknown dependencies Assess the situation using tools and interviews, and launch a pilot
Move everything like-for-like One path per workload; rehost only where speed matters
No one is responsible for the costs Tags and budgets from the landing zone, and a consistent FinOps rhythm
All at once (Big Bang) Waves with a fallback scenario
Team not involved Knowledge transfer and training throughout the process
Security after the fact Guardrails and logging in the landing zone, before the first workload

13. Frequently asked questions about Cloud Migration

All frequently asked questions

The terms are often used interchangeably. Cloud transition typically refers to the broader change—encompassing not only technology but also processes, management, costs, and your team’s expertise. Cloud Migration refers to the migration of applications, data, and the infrastructure itself.

You can migrate a single application in a matter of weeks. A data center with hundreds of servers is migrated in phases, often spread out over a year or longer. Gartner estimates that a complete migration away from VMware will take three years or more. A thorough assessment beforehand ensures reliable planning.

Not necessarily. Workloads with fluctuating demand are often cheaper in the cloud, while workloads that run 24/7 at a constant load are sometimes more expensive. The difference lies in rightsizing, reservations, shutting down what you’re not using, and factoring in management and hardware replacement costs.

Moving workloads back from the public cloud to an in-house data center or private cloud. Organizations do this for reasons of cost, sovereignty, or performance. Just as with migration, you make that decision on a per-workload basis: if something works well in the cloud, you leave it there.

Well-known European providers include STACKIT, OVHcloud, IONOS, and Scaleway. In addition, hyperscalers offer sovereign options, such as the AWS European Sovereign Cloud. A European provider operates under European jurisdiction but offers fewer managed services and less tooling. You choose based on each dataset.

In a hybrid cloud, you connect a private environment—whether on-premises or hosted by a provider—to one or more public clouds, sharing identities, networking, and management. With multicloud, you use multiple public clouds side by side, usually for a service that one provider delivers best. As a general safeguard against lock-in, multicloud is expensive. SUE keeps workloads portable using Infrastructure as Code and containers, where it’s worth the effort.
On 11 October 2027, Broadcom will discontinue general support for vSphere 8 and ESXi 8. Your options are listed on the vSphere 8 end of life page.
An exit plan describes, for each workload, how you will leave a cloud provider: which services you will replace, which dependencies you will intentionally retain, how you will export data and in what format, and what deadlines and costs are specified in the contract. Your infrastructure is defined in code, so you can rebuild it elsewhere. Test the plan at least once a year by restoring a workload in a different environment. SUE incorporates the exit plan into the migration roadmap of the Cloud Readiness Assessment.

Which workloads are ready for the cloud?

You can determine which workloads are ready for the cloud by conducting a Cloud Readiness Assessment.

Schedule a Cloud Readiness Assessment · Cloud Migration at SUE

Jonah Sanderse
Jonah Sanderse

Let’s talk


Which workloads are ready for the cloud?

You can determine which workloads are ready for the cloud by conducting a Cloud Readiness Assessment.

Schedule a Cloud Readiness Assessment · Cloud Migration at SUE

* required

We will respond within one business day. By submitting this form, you confirm that you have read and understood our privacy policy.
SUE Cloud & AI Native Solutions
Privacy overview
This website uses cookies. We use cookies to ensure that our website and services function properly, to gain insight into the use of our website, and to improve our products and marketing. For more information, please read our privacy and cookie policy.