New series · Data Architecture
Long-form, research-backed guides on data architecture, data engineering, governance, analytics, machine learning and generative AI — from ingestion to agents.
Press Ctrl K to search every article.
What data architecture really is, the six layers of a modern platform, how we got from warehouses to open lakehouses, the patterns that matter in 2026 — and a practical framework for choosing yours.
Kimball, Data Vault 2.x, One Big Table and medallion layers compared on one example, plus a layered recipe for a lakehouse people and AI can trust.
Why the catalog is the most strategic lakehouse decision: how the Iceberg REST protocol works, how Polaris, Unity, Glue and Horizon compare, and how to choose.
How Iceberg, Delta Lake, Hudi and DuckLake turn Parquet files into reliable tables — metadata, commits, deletes, maintenance — and how to choose in 2026.
Warehouse, lake or lakehouse? How each really works, why the lines blurred in 2026, and a decision tree, scorecard and scenarios to choose yours.
Why this blog exists, what it will cover, and how the series are organized — from data architecture to AI agents.
Series roadmap
From first principles to reference architectures: how modern data platforms are designed, why they look the way they do, and how to make the big decisions.