Stackable Docs Hub

Stackable

Stackable

How do you create acceptance criteria for a sovereign data platform?

Defining what „sovereignty“ means for a data platform is straightforward in principle. Translating that definition into concrete, testable acceptance criteria is where most organisations run into trouble. The gap between a vendor’s sovereignty claims and what a platform actually delivers in practice can be significant, and discovering that gap after procurement is expensive in every sense of the word.

This article walks through how to build acceptance criteria for a sovereign data platform systematically, from understanding what sovereignty genuinely demands, through mapping regulatory requirements, to turning your criteria into a repeatable evaluation framework. We look at how the Stackable Data Platform (SDP) approaches these requirements specifically at the end.

What sovereignty actually demands from a data platform

Data sovereignty is not a feature. It is a set of conditions that must hold simultaneously across infrastructure, governance, and operations. A platform is sovereign when the organisation running it retains meaningful control over where data lives, who can access it, how it is processed, and what software is executing on it at any given time.

That last point is often underestimated. Software supply chain transparency is a core sovereignty requirement, not a security add-on. If you cannot verify what is running in your environment, you cannot make credible claims about data governance. Sovereignty also demands that the platform does not create structural dependencies that force future decisions, whether those are licensing dependencies, proprietary API dependencies, or operational dependencies on a single vendor’s support organisation.

In practical terms, this means a sovereign data platform must run in infrastructure you control, use open and interoperable standards, provide full visibility into its software components, and allow you to exit without catastrophic migration costs. These are the foundational demands that acceptance criteria must reflect.

Mapping regulatory requirements to platform criteria

Regulatory frameworks give sovereignty requirements legal weight, and they are increasingly specific about what platforms must demonstrate. In European contexts, the Digital Operational Resilience Act (DORA) and the NIS-2 Directive both impose obligations that translate directly into platform evaluation criteria.

DORA, which applies to financial entities and their critical ICT providers, requires demonstrable control over third-party technology risk. That means your data platform must support contractual exit strategies, provide audit trails for data processing, and allow you to test resilience without depending on vendor cooperation. The NIS-2 Directive extends similar requirements to a broader set of critical infrastructure operators, with explicit attention to supply chain security.

Mapping these requirements to platform criteria means asking specific questions. Can the platform produce a software bill of materials (SBOM) for every component? Does it support policy-as-code for access governance? Can it be operated entirely within your own infrastructure without phoning home to a vendor’s control plane? Each regulatory obligation should generate at least one testable criterion.

Core acceptance criteria every sovereign platform must meet

With the regulatory and conceptual groundwork in place, you can define a baseline set of acceptance criteria that any sovereign data platform must satisfy. These are not aspirational goals; they are pass/fail conditions.

Infrastructure and deployment control

The platform must be deployable in infrastructure you own or control, including on-premises data centres, private cloud environments, and hybrid configurations. It must not require persistent connectivity to a vendor’s external services for normal operation. Kubernetes-native platforms have a structural advantage here because Kubernetes itself is a vendor-neutral, open standard with a broad ecosystem of conformant distributions.

Software transparency and supply chain integrity

Every component must be identifiable, versioned, and verifiable. The platform should support SBOM generation and ideally align with Supply-chain Levels for Software Artifacts (SLSA) standards. Open-source licensing must be clear and compatible with your organisation’s policies. Proprietary components embedded in otherwise open platforms are a common sovereignty risk that evaluation criteria must explicitly surface.

Governance and access control

The platform must support fine-grained, auditable access control that integrates with your existing identity infrastructure. Policies should be expressible as code so they can be version-controlled, reviewed, and tested. Federated governance, where different teams or business units manage their own data domains within a shared policy framework, is increasingly a requirement for organisations adopting data mesh architectures.

Vendor independence and portability

Data formats, APIs, and operational tooling must use open standards. The cost and complexity of migrating away from the platform should be assessable before adoption, not after. This criterion is often the hardest to evaluate because vendors rarely volunteer information about exit complexity.

How to validate criteria across technical and organisational layers

Defining criteria is only useful if you can actually test them. Validation needs to happen at two levels: technical and organisational.

At the technical level, validation means running the platform in a representative environment and verifying each criterion directly. Can you deploy it without external connectivity? Generate an SBOM? Apply a policy change through code and verify it takes effect? These tests should be scripted and repeatable, not one-time manual checks. A data security demo environment can be a useful starting point for understanding how access control and governance behave under realistic conditions.

At the organisational level, validation means checking whether the vendor’s commercial model supports sovereignty in practice. Does the support contract require you to run specific vendor-managed components? Are security advisories published openly, or do you need a support subscription to receive them? Is the roadmap publicly visible? Organisational validation often reveals sovereignty gaps that technical testing misses.

Common gaps when evaluating data platform sovereignty

Several gaps appear consistently when organisations evaluate data platform sovereignty, and they are worth naming explicitly so they do not catch you by surprise.

The first is conflating open-source licensing with sovereignty. A platform can be open-source licensed and still create significant lock-in through proprietary operational tooling, managed-only deployment models, or data formats that are technically open but practically difficult to migrate. Open-source is a necessary condition for sovereignty, not a sufficient one.

The second gap is ignoring the operational layer. Many evaluations focus on the data plane, where data is stored and processed, and overlook the control plane, where the platform is configured, monitored, and updated. If the control plane depends on vendor-managed infrastructure or proprietary tooling, sovereignty claims about the data plane are weakened considerably.

The third gap is treating sovereignty as a point-in-time assessment. Platforms evolve, licensing terms change, and vendor ownership structures shift. Acceptance criteria should include ongoing conditions, such as continued open-source licensing and publicly accessible security advisories, not just initial deployment requirements.

Turning acceptance criteria into a repeatable evaluation framework

A list of criteria becomes a framework when it is structured, weighted, and reusable across evaluations. The goal is to make future platform decisions faster and more consistent, not to create a one-time checklist.

Start by grouping criteria into categories that reflect your organisation’s priorities: infrastructure control, software transparency, governance capability, vendor independence, and operational maturity are reasonable starting categories. Within each category, distinguish between mandatory criteria (hard requirements that eliminate a platform from consideration) and weighted criteria (factors that differentiate between platforms that pass the mandatory threshold).

Document the evidence required for each criterion. „Supports open standards“ is not testable as written. „Stores data in Apache Iceberg format with no proprietary extensions required for read access“ is. Specificity is what makes criteria actionable rather than aspirational.

Finally, assign ownership. Technical criteria need engineers to validate them. Contractual and organisational criteria need procurement and legal involvement. A framework that assigns validation responsibility alongside each criterion is far more likely to be used consistently than one that leaves ownership ambiguous.

How Stackable supports sovereign data platform requirements

The SDP is built around the conditions that acceptance criteria for sovereignty typically surface. Here is what that looks like concretely:

  • Fully open-source and cloud-agnostic: Every component of the SDP is open-source, with no proprietary control plane or mandatory vendor-managed services. It runs on any Kubernetes-conformant distribution, on-premises or in any cloud.
  • Kubernetes-native operators for open-source data apps: The SDP provides operators for Apache Kafka®, Apache Spark™, Trino, Apache NiFi, Apache Airflow, and others. Each operator manages the full lifecycle of its respective application within your infrastructure. The Stackable Operator for Apache Kafka® is one example of how this works in practice.
  • Policy-as-code governance: Access control and data governance policies are expressed as code, version-controlled, and applied consistently across the platform, supporting federated governance models.
  • Software supply chain transparency: Stackable maintains a fully traceable software supply chain, with clear component versioning and open security advisories.
  • No vendor lock-in by design: Data formats and APIs follow open standards. The Data Lakehouse demo using Apache Iceberg, Trino, and Apache Spark™ illustrates how open interoperability works across the stack.
  • Fair, transparent pricing: Community edition access is free. Commercial support subscriptions are available for organisations that need them. See the plans page for details.

If you are working through a data platform evaluation and want to test these criteria against the SDP directly, get in touch with the team to discuss your specific requirements.

Ähnliche Artikel

Comments are closed.