๐Ÿ”ง Chapter 07: Platform Comparison

Apache Atlas

Type: Open Source | Best For: Hadoop Ecosystem

Pros: Tight Hadoop integration, strong type system, free

Cons: Complex setup, dated UI, limited cloud support

DataHub

Type: Open Source | Best For: Cloud-Native, Modern Stack

Pros: Modern architecture, 100+ connectors, active community, GraphQL API

Cons: Relatively new, learning curve

Architecture: React UI โ†’ GraphQL โ†’ Spring Boot โ†’ MySQL/Postgres/ES/Neo4j

Collibra

Type: Commercial | Best For: Enterprise, Strong Governance

Pros: Comprehensive features, professional support, regulatory compliance

Cons: High cost, complex setup, limited customization

Amundsen

Type: Open Source | Best For: User-Friendly Discovery

Pros: Excellent UX, easy to use

Cons: Limited governance features

Alation

Type: Commercial | Best For: Collaboration, Data Culture

Pros: Crowdsourcing, strong collaboration features

Cons: Expensive

Selection Guide

Reference Standards

Governance is the difference between a catalog that drifts into uselessness and one that compounds in value. WIA-DATA-007 governance follows:

Roles and Responsibilities

The reference model assigns five roles using RACI semantics:

Policy Enforcement

Each catalog asset carries machine-readable policy attributes (sensitivity_class, retention_days, allowed_jurisdictions, license_id) so downstream systems can enforce policy automatically. Policies are versioned, and the catalog records every policy change as a PROV-O activity for audit purposes.

Practical Application

Governance lives in policy, but it dies without instrumentation. Reference deployments treat governance as a feedback loop:

  1. Policy as data. Every policy is encoded as machine-readable attributes on each asset (sensitivity_class, retention_days, allowed_jurisdictions, license_id), so downstream systems can enforce automatically rather than via PDF circulation.
  2. Reviews on a calendar. Quarterly access reviews per ISO/IEC 27001 Annex A and annual data inventory reviews per ISO/IEC 27701 are scheduled events on the catalog itself, with completion telemetry visible to the data protection officer.
  3. Incidents become improvements. Each governance incident produces a closed-loop entry: what attribute was missing, which policy was unclear, who is updating which template. The catalog's governance dashboard is, in effect, the management system required by ISO/IEC 38505-1.

This loop, far more than any single policy document, is what makes governance defensible to an external auditor.

WIA-DATA-007 โ€” Conformance with ISO/IEC 11179, W3C DCAT v3, ISO/IEC 25012, FAIR principles. Citations per CITATION-POLICY.md ยง2.1 ALLOW.