Architecture Patterns Deep Dive

Part 2: The Architecture Landscape

Part 1 built the vocabulary and physics an architect reasons with — ACID and CAP, OLTP vs OLAP, batch vs NRT vs RT. Part 2 turns that vocabulary into named, arguable shapes: the six patterns in this group are the actual whiteboard sketches an architect defends in a design review, each one a specific bet about how to reconcile latency, storage economics, curation, and ownership. Taken together, the seven topics here aren’t seven independent tools to memorize — they form a landscape with real structure: two answer how data gets processed (Lambda, Kappa), two answer how it’s stored and curated (Lakehouse, Medallion), two answer who owns it and how it’s found (Data Mesh, Data Fabric), and the closing topic shows how those three axes combine into one real system rather than a single pick from a menu.

mindmap
  root((Architecture Patterns Deep Dive))
    Lambda Architecture
      Batch + speed layers, reconciled at query time
    Kappa Architecture
      Stream-only, replay the log instead of a batch layer
    Lakehouse Architecture
      Open table formats add ACID to object storage
    Medallion Architecture
      Bronze/Silver/Gold, trust level legible by layer
    Data Mesh
      Domain ownership, data as a product
    Data Fabric
      Active metadata, automated discovery across a heterogeneous estate
    Choosing Among the Patterns
      Decision matrix across all six

See also: Lakehouse Architecture: Unifying Warehouse & Lake connects to Table Formats: Delta vs Iceberg vs Hudi in the Storage & Table Formats group, which goes deeper on how Delta, Iceberg, and Hudi differ.

Topics

#	Topic
1	Lambda Architecture: Batch + Speed Layers
2	Kappa Architecture: Stream-Only Processing
3	Lakehouse Architecture: Unifying Warehouse & Lake
4	Medallion Architecture: Bronze/Silver/Gold
5	Data Mesh: Decentralized Domain Ownership
6	Data Fabric: Metadata-Driven Integration
7	Choosing Among the Patterns: Comparison & Decision Guide

---

| ← Previous: Choosing an Architecture & the Road to Becoming a Data Architect | Next: Lambda Architecture: Batch + Speed Layers → | |:—|—:|

---

Table of contents

• Lambda Architecture: Batch + Speed Layers
• Kappa Architecture: Stream-Only Processing
• Lakehouse Architecture: Unifying Warehouse & Lake
• Medallion Architecture: Bronze/Silver/Gold
• Data Mesh: Decentralized Domain Ownership
• Data Fabric: Metadata-Driven Integration
• Choosing Among the Patterns: Comparison & Decision Guide