← Back to principles
Data
Right tool for the data job
Statement
Data storage and tooling should be fit for purpose based on data shape and usage patterns
What this means in practice
Data comes in many forms -- structured, semi-structured, documents, video, geospatial, vector -- and the way it is consumed varies dramatically across transactional processing, analytics, caching, and search. In each case, the storage technology and data modelling approach should be selected to match the shape of the data and its intended use, rather than forcing everything into a single platform. This applies not only to database engine selection but also to how models are organised, such as star schemas in analytical contexts versus normalised models in transactional systems.
However, fitness for purpose must be balanced with operational realism. Every tool must satisfy universal core requirements around security, availability, scalability, and recoverability. Teams must also consider the operational burden; expecting a client to maintain in-house expertise across many different database technologies is not sustainable.
Why this matters
Using the wrong tool for a given data problem leads to poor performance, unnecessary complexity, and fragile workarounds. Equally, defaulting to a single technology for all use cases creates bottlenecks and compromises. Selecting the right tool, whilst keeping the operational footprint manageable, delivers better outcomes for both delivery teams and the organisations that maintain the solutions long term.
Practices that meet this principle
Evaluate data shape, volume, and access patterns before selecting storage technology
Document the rationale for technology choices in architecture decision records
Assess operational impact including skills requirements, backup and recovery, and monitoring needs
Ensure all selected tools meet baseline non-functional requirements for security, availability, and scalability
Use appropriate modelling techniques for the context, such as normalised models for OLTP and dimensional models for analytics
Review technology choices at key milestones as requirements evolve
Validation
A project meets this principle when:
Storage technology selection is documented with rationale linked to data shape and usage patterns
Non-functional requirements are assessed and met for each selected technology
OR:
No storage technology has been selected without documented justification