Architectural Principles
Architectural principles influence and shape the way you design and deliver a SATRE-aligned TRE. They are a set of guiding considerations that sit above any specific architectural requirement, and can be applied across the entire architecture.
They consist of the following parts:
- Statement:
A singular sentence that summarises the principle
- Rationale:
Justification as to why this principle is important for the specification
- Implications:
Things you need to consider or do to practise this principle
Principle 1 - Usability
Statement
A TRE instance that works for all users minimises barriers to use, and provides a productive and accessible analysis environment for research.
Rationale
There is often a trade-off between increased operational security and the usability of a TRE. In order to maintain productivity, a TRE must balance these two competing aims. The design and configuration of a TRE should allow all individuals involved with a TRE to effectively fulfil their roles.
Implications
Robust TRE design and implementation should start by understanding users’ diverse expectations, needs, existing skillsets and preferences and responsibilities.
Design, configuration and testing of TREs must recognise a diversity of users. For instance, not all users are researchers and not all researchers are users. Other users include TRE operators, information governance officers, and TRE builders/developers.
Because of diverse user needs, it is unlikely that a specific TRE instance will perfectly match the needs of all users.
A TRE that is overly strict on tool and software provision may risk becoming unusable for users with different and varied backgrounds and skillsets.
Working environments can differ significantly from users’ preferred setups. This has design and resource implications for supporting new users, and consideration should be given to resources and time required to help users get up to speed with new and unfamiliar TRE instances.
Improving user experience takes time and resource, and will involve trade-offs between investing time in improved standards, better functional design, improving work and organisational culture, boosting users’ skills and knowledge through training and making help more readily available at an organisational level. These trade-offs will need to be addressed at an organisational level, and teams may want to consider resourcing staff to focus specifically on these questions, for instance in the positions of product managers or service functions.
Principle 2 - Maintaining public trust
Statement
TREs holding public data should build and maintain the trust of data subjects and any other individuals, groups, communities and organisations impacted by the research conducted using TREs. TREs should demonstrate trustworthiness by protecting privacy, keeping data secure and being transparent about their work.
Rationale
It is vital to maintain public trust in the ways TREs hold and use data so that this data can be used to deliver public benefits. This could include maintaining the trust of members of the public whose data is held, and those who are impacted by research conducted using TREs.
Trust is fragile and can be eroded by data breaches, opaque decision-making, or the perception that data is being used in ways that were not agreed upon or do not serve the public interest.
Public engagement work has indicated there is support for the use of regulated and ethical TREs working for the public benefit as long as conditions are met surrounding security and transparency. Consulting impacted parties, including the public, can help ensure TREs are being used for positive, impactful and agreed-upon purposes. Being transparent about the data held and the projects or organisations who access the data can also help maintain trust.
Implications
Transparency is key to building trust. TREs holding public data should be transparent about the TRE operations, data held and research projects accessing data. This information should be made publicly available in plain language and an accessible format.
Trust implications of research outputs, not only the data held need to considered. Publications, models, and derived datasets produced using TRE data can affect public trust particularly where outputs could be used to re-identify individuals, serve commercial interests, or are perceived as inconsistent with the purposes for which data access was granted.
Actively and effectively involving members of the public in oversight of TREs and their processes is important for accountability. When involving public members they must have decision or advisory roles where their input can make an impact to TRE operation or governance.
Public involvement and engagement takes time and resources. TREs holding public data should consider allocating specific staffing and budget to public involvement and engagement activities.
Members of the public expect public data to be used for public benefit. The process of reviewing requests to access public sector data, should include members of the public where possible, and follow agreed-upon governance to ensure projects using this data are in the public benefit, and provide clarity around any commercial access.
Feedback loops should be established as part of TRE processes for transparency and accountability
Principle 3 - Observability
Statement
Human initiated and automated processes resulting in change within the TRE should be observable.
Rationale
System/process observability is key to understanding whether your policies and controls are actually doing what is intended.
It also allows operators to continuously improve their systems and processes, measure their effectiveness, and identify the causes of incidents. Data can also be made available to other parties such as auditors, data subjects and data providers as part of the assurance process.
This applies to both technical systems and policies/processes.
Implications
In order to understand what is happening within the TRE, both automated and human initiated processes should generate sufficient data, for instance through audit logs. Any generated data should follow standards for provenance and transparency for audit trails.
Different levels of observability may be needed for different users. Any data collected from an observability perspective should consider the needs of those who will use it, and minimise collection accordingly.
There may be ethical and confidential issues to consider when implementing the observability principle.
Principle 4 - Standardisation
Statement
TREs should adhere to standards or well-known patterns wherever possible.
Rationale
Standardisation makes it easier to design, operate, use and understand TREs, and reduces duplication of work. This includes making TREs easier to use, deploy, and audit.
TREs should be built in such a way that they do not restrict or prevent interoperability where this may be desirable in future, by identifying and avoiding or removing barriers to interoperability.
Standardisation is also linked to the public trust principle, as a standard approach to TRE provision will make it easier for impacted parties to understand how their data will be used within TREs.
Implications
Owing to the broad definition of TREs, there are currently no technical or information governance standards focused on TREs. The SATRE specification has therefore been designed to help Developers and Operators with a variety of technical/policy requirements consider their options
TRE Developers and Operators should be prepared identify the technical standards that are appropriate for them to work towards meeting whilst developing or maintaining their particular TRE(s).
Standards that TREs adhere to may range in scope, including technical, operational and governance requirements.
TRE Developers and Operators should ensure that when they aim to meet multiple standards, those standards align with one another to ensure there is no contradiction in requirements.
There might be good reasons why any particular TRE does not possess one or more of the capabilities listed in this specification, but most TREs should aspire to meet them in the long-term.