Most enterprises assume they control their data platform. They chose the tools; they signed the contracts; they manage the clusters. But control is not the same as ownership, and the gap between the two tends to show up at the worst possible moment: during a migration, an audit, or a licensing renewal where the numbers no longer make sense.
Data sovereignty is not a checkbox. It is a structural property of how your platform is built, where it runs, who can inspect it, and what happens when you want to leave. The seven questions below are designed to surface the hidden dependencies that most architecture reviews miss. We look at what a genuinely sovereign data platform architecture looks like, and how the Stackable Data Platform (SDP) approaches this problem, specifically at the end.
The hidden control gaps most enterprises overlook
Control gaps rarely announce themselves. They accumulate quietly through reasonable-sounding decisions: choosing a managed service because it reduces operational overhead, accepting a proprietary connector because it was the fastest path to production, or skipping the licence audit because the renewal seemed straightforward.
The most common gaps fall into a few categories. First, there is operational opacity: you run the software, but you cannot inspect how it behaves internally, what telemetry it sends, or how upgrades are applied. Second, there is portability debt: your data is technically yours, but it lives in formats or locations that make moving it prohibitively expensive. Third, there is licensing exposure: the terms that governed your deployment when you started may not be the terms that govern it today. Open-source licences have changed for several widely used data tools in recent years, and not always in the user’s favour.
These gaps are not always the result of bad decisions. They are often the result of decisions made without full information about long-term implications.
7 questions that reveal who truly owns your data stack
Work through these questions honestly. They are not a marketing quiz; they are a structural audit. Score one point for each „yes“ answer.
- Can you inspect the full source code of every component in your data platform? If any component is closed-source or obfuscated, you are operating on trust, not verification.
- Can you move your entire data platform to a different cloud provider or on-premises environment without rewriting pipelines? Cloud-native does not mean cloud-agnostic. Many managed services bind you to specific APIs, storage formats, or networking constructs.
- Do you control the upgrade schedule for every component? Automatic updates applied by a vendor without your approval are a form of operational dependency, even if the intent is helpful.
- Can you reproduce your entire platform configuration from code? If your platform state lives only in a UI or a vendor’s control plane, you do not fully own it.
- Are your data formats and APIs based on open standards? Proprietary formats create exit costs that compound over time. Open formats like Apache Iceberg keep your options open.
- Do you know exactly which third parties can access your data, and under what conditions? This includes the vendor’s support team, telemetry endpoints, and any sub-processors named in the contract.
- Could you continue operating your platform if your current vendor ceased to exist or changed their licensing terms tomorrow? This is the ultimate sovereignty test.
How scoring your answers exposes real dependency risk
The score itself matters less than the pattern of your answers. A score of seven means your platform is genuinely sovereign. A score of four or five likely means you have made reasonable trade-offs, but you should understand exactly where the exposure sits. A score below four suggests that significant structural dependencies exist, and that they may be invisible in your current risk register.
Pay particular attention to questions two and seven. Portability and survivability are the two dimensions that most directly determine whether data sovereignty is real or theoretical. An organisation that scores „yes“ on everything else but „no“ on those two has built a well-managed dependency, not an independent platform.
The questions around configuration reproducibility and open standards (four and five) are often where technically sophisticated teams are surprised. It is entirely possible to run open-source software in a way that creates proprietary lock-in, if the operational layer is opaque or the configuration is not managed as code.
Where proprietary platforms lose you control
Proprietary data platforms typically offer a compelling initial experience: fast setup, integrated tooling, and a single support contract. The control trade-offs become visible later.
The most significant is licensing risk. When a vendor changes the licence terms of software you depend on, your options are limited: accept the new terms, negotiate (from a weak position), or migrate under time pressure. Several major data infrastructure projects have undergone licence changes in recent years, moving from permissive open-source licences to source-available or commercial-only models. Teams that built on those projects without a portability strategy found themselves in difficult positions. If you are evaluating alternatives, understanding how to avoid vendor lock-in when choosing a platform is a critical first step.
The second is format and API coupling. Proprietary query engines, storage formats, and connector ecosystems create switching costs that are not always visible at procurement time. By the time the cost becomes apparent, the data volume and pipeline complexity make migration a multi-year project.
The third is operational opacity. When you cannot inspect how a platform component behaves, you cannot fully audit it, cannot verify its security posture independently, and cannot adapt it to your specific compliance requirements. For organisations subject to regulatory oversight, this is not just an inconvenience.
What a sovereign data platform architecture looks like
A genuinely sovereign data platform has four structural properties: it is open, portable, reproducible, and composable.
Open and inspectable
Every component’s source code is available for inspection, modification, and redistribution under a licence that does not change unilaterally. This is not just about cost; it is about the ability to verify behaviour, audit security, and adapt the software to your environment.
Portable across environments
A cloud-agnostic data platform runs consistently whether deployed on-premises, in a public cloud, or in a hybrid configuration. Kubernetes provides a consistent operational substrate that makes this achievable without maintaining separate deployment codebases for each environment. When evaluating your options, understanding how to choose between cloud and on-premises during migration can help you make the right architectural decision.
Reproducible from code
Infrastructure as code is not optional in a sovereign architecture. Every component, every configuration, every access policy should be expressible as code that can be version-controlled, reviewed, and reapplied. This is what makes a platform traceable and auditable.
Composable and modular
A modular architecture lets you add or replace components without rebuilding the entire platform. This is what keeps your options open over time. If a better query engine emerges, or if a component’s licence changes unfavourably, you can swap it out without a full migration.
Regulatory compliance and data sovereignty in European markets
For organisations operating in Europe, data sovereignty is increasingly a regulatory requirement, not just an architectural preference. The Digital Operational Resilience Act (DORA), which applies to financial entities, requires demonstrable control over ICT systems and third-party dependencies. The NIS-2 Directive (which expanded the scope of the original NIS Directive significantly) places obligations on critical infrastructure operators around supply chain security and incident reporting.
Both frameworks, as we understand them, push in the same direction: organisations need to know what software they are running, where their data is processed, and who has access to it. A platform built on closed-source components or managed entirely by a third party makes these obligations harder to meet, not because the vendor is necessarily untrustworthy, but because the verification mechanisms are limited.
The ability to run a fully traceable, auditable data platform in your own infrastructure, with open-source components whose behaviour you can inspect, is directly relevant to demonstrating compliance with these frameworks. For a deeper look at how open-source platforms address these obligations, see how to meet GDPR and data sovereignty requirements on an open source platform. „Made in Europe“ is not just a marketing phrase in this context; it reflects a set of legal and operational guarantees about where data is processed and under which jurisdiction.
Practical steps to reclaim control of your data infrastructure
If your audit score revealed gaps, the path forward is incremental, not a full rebuild. Start with visibility before attempting migration.
- Map your dependencies explicitly. Document every component, its licence, its data access scope, and what happens to your platform if that component becomes unavailable or changes terms.
- Prioritise open data formats. Migrating to open table formats like Apache Iceberg for your data lake layer is one of the highest-leverage steps toward portability. It decouples your data from any specific query engine.
- Move configuration into code. If your platform state is not fully reproducible from a version-controlled repository, that is the first operational gap to close. Kubernetes-native platforms make this significantly more tractable.
- Audit third-party access. Review your contracts and data processing agreements to understand exactly which parties can access your data, under what conditions, and with what notice requirements.
- Test portability before you need it. Run a tabletop exercise: what would it take to move your platform to a different environment in 90 days? The answer will tell you more about your actual sovereignty than any vendor questionnaire.
None of these steps require a complete platform replacement. They are about building the structural properties of sovereignty incrementally, so that when you do need to exercise control, the mechanisms are already in place.
How Stackable helps with data sovereignty
The SDP is built specifically around the architectural properties described above. Here is what that means in practice:
- Fully open source: Every component of the SDP is available under an open-source licence. The source code is publicly accessible, and the platform is developed transparently. There are no hidden components or proprietary extensions required for production use.
- Kubernetes-native and cloud-agnostic: The SDP runs on any Kubernetes cluster, whether on-premises, in any public cloud, or in a hybrid configuration. There is no dependency on a specific cloud provider’s APIs or managed services.
- Infrastructure as code by design: Platform configuration is managed declaratively through Kubernetes custom resources. Every component, every access policy, every configuration change is version-controllable and reproducible. The Data Lakehouse demo shows this in practice with Apache Iceberg, Trino, and Apache Spark™ working together in a fully declarative setup.
- Modular and composable: The SDP includes operators for Apache Kafka®, Trino, Apache Spark™, Apache NiFi, and other components. Each operator can be deployed independently. You add what you need and remove what you do not, without rebuilding the platform.
- Designed for European regulatory requirements: The SDP is developed in Germany, with data sovereignty and regulatory compliance as first-class design considerations, not afterthoughts.
If you want to see where your current platform stands against these criteria, or explore what a sovereign architecture looks like for your specific environment, get in touch with the Stackable team. We are happy to work through the specifics without a sales pitch attached.
Ähnliche Artikel
- How does Apache NiFi handle data ingestion in a lakehouse?
- What is an open table format and why does it matter for data lakehouses?
- What is the total cost of ownership of a data lakehouse?
- How do you manage rollback risk during a data platform migration?
- What is the impact of a data platform migration on downstream BI tools?