How To Actually Make Decisions When Architecting a Data Platform
summary
Claims a repeatable decision process: version-controlled Non-functional requirements alongside functional specs, use Architectural Decision Records to record why choices are made, and prefer incremental upgrades and small PoCs over big-bang migrations. Demonstrates an evaluation workflow: assemble cross-functional reviewers, shortlist candidates, score features and vendor dimensions on a -2 to 2 matrix, unpivot the results in a notebook and visualize in Power BI to produce heatmaps and weighted rankings, then run targeted PoCs and pick the top product.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:10Data platform options
- 3:38Capturing requirements
- 5:04Organizational alignment
- 8:12Evaluating vendors
- 10:59Building evaluation team
- 13:06Structured evaluation
- 15:15Proof of concept tests
- 16:17Architectural decision records
- 17:42Final recommendations
related talks
25mThinking Architecturally
47mMaking architecture decisions, the Agile Way
40mEvolutionary Architecture - The art of making decisions
41mArchitecture decision making
39mOne Does Not Simply Make Architecture Decisions Alone: ADRs and Architec
19mHow to Architect your Data Platform for Success
46mCreating Effective and Objective Architectural Decision Records (ADRs)
41mBeyond "it depends" - How much architectural work do we need?