GDS Data Architecture

• About

Data Sharing Principles

• Overview
• 1. Treat data as an asset
• 2. Federate first
• 3. Reuse sharing solutions
• 4. Support automation
• 5. Design for all data stakeholders
• 6. Use common standards for sharing
• 7. Share data transparently
• 8. Share data lawfully and ethically
• 9. Secure shared data proportionately

Capability Model

• Overview
• Channels
• Infrastructure
• Data Access
• Services
• Governance
• Security
• Trust

2. Federate first

Directional principle

Statement

We consider federated approaches to data sharing first, where data remains with its owners and code moves to the data.

Why does this matter?

Working towards a joined up digital public sector doesn’t require the mass movement of data assets or long battles over decision-making authority. Instead, we should aim to implement federated approaches to data sharing that provide a unified view of data from multiple sources without moving or copying it. This will allow decisions over data to always remain close to those who control it and understand it best.

Federated approaches have multiple advantages over more centralised or informal data sharing methods. They create a strong foundation for responsible data use by encouraging formal expertise sharing and limiting third party interference. They are inherently scalable thanks to their decentralised nature. Finally, organisational working models can be better preserved when every actor has full and equal autonomy over their data sharing decisions.

How do we do this?

Overall, we should make data governance and architecture provisions for federated data sharing solutions and reduce movement of our data assets.

Data providers should

• Work with data sharing enablers to create technical and governance structures that support federated models of data sharing.
• Invest time and resources to build capacity for operating new technical and governance structures.
• Avoid sharing copied data and justify any data copying with a clear business or legal need.

Data consumers should

• Work with data sharing enablers to ensure that suitable analysis and processing capabilities are available through federated solutions.
• Ensure that code used for processing shared data will be legible for interfaces close to the source data.
• Use and request access to data sharing services before considering requests for copied data.

Data sharing enablers should

• Identify modern digital architecture designs and patterns that could support federated data sharing within the context of their work.
• Work with data providers to create technical and governance structures that support federated models of data sharing.
• Consider how data analysis and processing capabilities could be deployed close to data sources through APIS or other interfaces.