Sovereign cloud: control access, data and operations

Share this article

Who can keep your cloud running?

Consider this example: your IT provider becomes unavailable. Your website still runs, but nobody in the business can manage the domain, retrieve a backup or restore the database. Your company pays the bill. The provider still holds the access credentials.

This gives cloud sovereignty a practical meaning: which decisions can you make, what information can you access, and who can take over operations? A European server location answers only part of that question. Contracts, administration rights, subprocessors and the underlying technology also need review.

This overview covers control during everyday operations. The separate cloud exit guide covers a planned migration; the vendor lock-in article covers choosing a new service.

Review dependencies for each service

Vendor lock-in doesn't happen overnight. It starts with a handy serverless function here and a proprietary "fully managed" database there. But every line of code tied to vendor-specific APIs (like AWS SQS or Google Cloud Spanner) is a tether that limits your future moves.

Review dependencies for each service: who controls the customer account, which data can be exported, and which features would need replacing? An intentional dependency can be worthwhile when a service saves your team work and the consequences of switching are understood.

The Data Act includes rules for switching between data processing services. It does not create a general obligation for SMEs to use European providers or a particular architecture. Legal requirements depend on the service, the data and the company’s role.

Compare cost and operating effort

Running costs and switching charges

Compare the total cost: usage, administration, backups, support and a possible switch. A low server price can be offset by additional operating work. A more expensive managed service can be economical when it takes on tasks your team would otherwise need to perform.

Switching charges within the Data Act’s scope are due to be removed on 12 January 2027. That does not make the whole migration free: application changes, data validation and running both environments can still require work.

Dependencies inside the application

When an application uses provider-specific APIs, identify the affected code and available alternatives. Estimate the work after that review. A fixed percentage of the entire codebase tells you little about your system.

PostgreSQL and containers do not guarantee an unchanged move either. Versions, extensions, permissions, networking and connected services need to work at the destination. A successful trial using a data copy provides better evidence than the product name alone.

Diagnostic: How Sovereign is Your Cloud?

Start with one business-critical service and review these questions with its operator:

Question Useful evidence
Who administers the main account? A company-controlled account with a documented deputy
Can the data be used elsewhere? An export and verified import of a representative data copy
Who can take over after an outage? A rehearsed recovery procedure and available access
Which dependencies remain? A list of APIs, licences and external services

An unanswered question becomes a concrete task. Prioritise it by the impact of failure: an untested restore of order data matters more than exporting a rarely used utility. Assign a person and a next review date to each task.

Infographic comparing the Proprietary Trap with the Sovereign Cloud architecture
The Sovereign Cloud relies on standardized layers (K8s/Docker) and open-source tools to make dependencies visible and prepare for a move.

Architecture for the Sovereign Cloud

Document applications, data and connected services separately. For each component, identify what depends on the current provider and which alternatives could work.

What containers make portable

Kubernetes provides a shared model for running containers. However, components such as StorageClasses and their storage providers may have different configurations. Document those dependencies and check them in the target environment. For a small website, a simpler server setup may be more appropriate than running a cluster.

Run n8n and Postgres yourself

Running your own n8n instance gives you more control over configuration and storage. You also take responsibility for updates, access and recovery. The n8n self-hosting documentation assumes the necessary operating skills. Connected APIs can still send data to external providers; self-hosting alone therefore does not establish GDPR compliance.

Check whether someone else can take over

Ask a second qualified person to follow your documentation through a low-risk part of operations, such as restoring a test copy. They should be able to identify the backup, the required access and the criteria for a usable result. Production records and production credentials do not belong in a freely accessible exercise document. Use a separate environment and the permissions intended for that work.

Start with the company account. Record who can administer it and who provides cover. Check that invoices, contract changes and security notices reach a monitored business address. An account that can only be recovered through a former employee’s personal telephone number is difficult to take over during an incident. A password list alone does not resolve that organisational dependency. The recovery arrangements need to remain available when the usual contact is absent.

Next, examine the data. Ask the operator to demonstrate an export and explain its contents. Customer information may be distributed across tables, file storage and connected systems. A contact export might omit attachments or links to orders. Compare a small sample in the source and restored environment. Record missing fields, permissions and relationships before describing the export as complete. The sample should represent the work the business actually needs to perform.

The application itself needs configuration records and operating instructions. Include external services: which system sends email, which interface supplies product information, and which domain points to the application? Provide confidential values through the approved secure access mechanism; the instructions should explain where they are needed. They should also describe how to recognise a failure and who decides what happens next. Someone taking over needs to understand those dependencies before changing the system.

Handover area What the second person should demonstrate
Account and contract Responsibility, cover and a monitored contact address
Data Export scope and a sample check after restoration
Application Startup in the test environment and a critical workflow check
Failure handling Stop condition, escalation and return to the previous state

Afterwards, discuss where the second person needed help. Those questions reveal missing instructions or knowledge that the original operator took for granted. Update the documentation where each question arose. A later exercise should be possible without relying again on the original administrator’s memory. Choose the scope according to the impact of the service involved, and retain the outcome so the next person knows what was actually demonstrated.

Decide which responsibilities should stay with the provider

Running everything yourself transfers work to your team. Someone needs to prepare updates, investigate failures, check backups and provide cover during absences. Review the control you actually need for each service. For a replaceable utility, a documented export may be sufficient. For an application used to process daily orders, a tested recovery procedure and an agreed handover to another operator may also be necessary. These are different operating commitments.

Record the decision in a few sentences: what value does the service provide, which dependencies do you accept, and what would trigger another review? A price change, a different processing location or the removal of an important feature might be such a trigger. Future staff can then understand why the service was chosen without reconstructing every earlier conversation. This also gives procurement a usable starting point before a contract comes up for renewal.

When reviewing an offer, ask for a precise division of tasks. “Managed” can cover different services. Establish who maintains the operating system, application, database and backups, and who performs a recovery attempt. Clarify what your own team must contribute and which work requires a separate order. A readable responsibility record helps both daily cooperation and a later handover. It also gives you a specific question to ask when an unexpected task falls between two providers.

A system change is a useful reason to revisit these records. A new API can alter the data flow. Replacing a service may invalidate export instructions. Record the date of the last actual check and any known changes since then. A documented gap helps the next decision more than an old confirmation whose assumptions no longer hold. Decide who updates the record as part of the change, while the relevant information is still available.

Technical support from Planfold

A first discussion can establish which service to review. Bring an overview of the relevant accounts, existing operating instructions and an example of a critical business workflow. These help identify missing information and useful technical work.

Before implementation, agree what will be handed over: documented configuration, an export procedure or recovery instructions, for example. Specify who accepts the result and who maintains it afterwards. If Planfold provides ongoing support, its scope, responsibilities and limitations are agreed separately. A later provider switch still requires its own plan.

Start with one service

Choose an important service and ask to see its main account, data export and recovery procedure. Record what works and which access details or instructions are missing. This gives your team a manageable first task.

If you then want to take on more responsibility, plan for operating time, cover during absences and the necessary skills. The platform ownership playbook explains how to divide that work.

Strategic takeaway model for digital sovereignty with solid foundations
Sovereignty is the freedom to choose the best partner for your next move.

Related Posts