Skip to content

What is Data Mesh?

Data Mesh is an approach to managing and sharing data across an organization that focuses on decentralized ownership, high-quality data products, and self-serve access. Instead of a single central data team controlling all data pipelines, each domain team is responsible for the quality, accessibility, and documentation of the data they produce.

When combined with Event Sourcing, Data Mesh principles help turn event streams into discoverable, reliable, and reusable data products.

The Core Principles

While Data Mesh can be described in many ways, four principles are essential:

  • Domain-oriented ownership: The team that owns the business domain also owns its data products.
  • Data as a product: Data is delivered with clear purpose, quality guarantees, and versioning.
  • Self-serve infrastructure: Teams can publish, discover, and consume data without bottlenecks.
  • Federated governance: Global policies and standards ensure interoperability without centralizing all decisions.

Why It Fits with Event Sourcing

Event Sourcing naturally supports Data Mesh ideas:

  • Immutable source of truth: Domain events form the foundation of consistent data products.
  • Clear semantics: Events use the language of the business, making data products easier to understand.
  • Versioned projections: Changes in schema or logic can be handled without losing historical context.
  • Reproducibility: Any data product can be rebuilt exactly as it was at a specific point in time.

From Events to Data Products

With Event Sourcing, the path to Data Mesh looks like this:

  1. Capture domain events in the event store.
  2. Build projections tailored for analytics, operational reporting, or machine learning.
  3. Publish projections as data products with documentation, SLAs, and versioning.
  4. Expose data products via APIs so other teams can consume them directly.

For example, a library might expose:

  • Borrowing history per member – for engagement analysis.
  • Title demand over time – for inventory forecasting.
  • Overdue patterns – for service improvement and targeted notifications.

Lightweight, Not Over-Engineered

Many Data Mesh discussions focus on complex organizational charts and heavy tooling. Here, the focus is simpler: provide high-quality, domain-owned data in a form that others can easily use.

Event Sourcing already gives you a strong foundation – all that's needed is a consistent way to package and share the projections.

Next up: AI, Machine Learning, and Data Science – how to use data products to turn facts into insights, predictions, and decisions.