The Core Principles of Data Mesh
Data Mesh rests on four interlocking principles. Understanding each one — and why it exists — is the foundation for knowing whether and how to implement it.
Each principle in Data Mesh addresses a specific failure mode of centralised data architectures. They are designed to work together — removing one breaks the system. A company that adopts domain ownership without self-serve infrastructure just creates decentralised chaos. A company that builds a self-serve platform without federated governance creates ungoverned sprawl. The four principles are interdependent by design.
Domain Ownership
The team closest to the data owns it.
Data is owned, managed, and served by the domain team that generates it — not by a central data team. The logistics team owns logistics data. The sales team owns sales data. This aligns accountability with expertise.
Benefits
- Faster data changes without central team bottlenecks
- Higher data quality because owners understand the context
- Clear accountability for SLAs and schema evolution
Watch out for
Requires domain teams to take on data engineering capabilities — tooling and training investment needed.
Data as a Product
Data outputs are first-class products.
Domain teams don't just expose data — they build data products with clear contracts, documentation, quality guarantees, and versioning. Consumers can discover and trust these products the same way they trust well-built software APIs.
Benefits
- Discoverable, trusted data assets across the organisation
- Stable interfaces that consumers can rely on
- Quality SLAs that align with business requirements
Watch out for
Shifts the mindset of data engineers from pipeline builders to product owners — a cultural change that takes time.
Self-Serve Data Infrastructure
Domain teams shouldn't need to be platform engineers.
A shared infrastructure platform provides the tooling domain teams need to build, deploy, monitor, and share data products — without requiring each team to build their own storage, compute, and governance stack from scratch.
Benefits
- Reduces infrastructure duplication across domain teams
- Lowers the barrier to entry for non-specialist teams
- Enables consistent operational standards across domains
Watch out for
The hardest principle to implement — requires significant investment in platform engineering upfront.
Federated Computational Governance
Global policies, local autonomy.
A global governance layer enforces organisation-wide policies — security, privacy, compliance, interoperability standards — automatically and programmatically. Domain teams operate freely within this envelope without being micromanaged.
Benefits
- Compliance and security enforced at scale without manual review
- Consistent data contracts and schemas across domains
- Audit trails and lineage maintained automatically
Watch out for
Encoding policies as code requires close collaboration between legal, security, and engineering teams.
Working with Enari
Enari supports organisations in designing and implementing Data Mesh architectures — from domain ownership workshops to self-serve platform engineering and federated governance frameworks. Talk to us about your data architecture →
Ready to start?
Let's build something exceptional
Tell us about your data challenge and we'll find the right solution together — fast, practical, and ROI-focused.