Skip to content

data-governance.md

Latest commit

 

History

History
129 lines (95 loc) · 3.97 KB

data-governance.md

File metadata and controls

129 lines (95 loc) · 3.97 KB
Raw

Data Governance Guide (v1.10.3)

This guide shows how to use OpenMetadata to implement practical data governance: ownership, roles and responsibilities, policies, domains, glossary/tags, SLAs, and stewardship workflows.

Related: Getting Started · Security & Compliance · Glossary

Last updated: October 29, 2025

Governance model

flowchart LR
	Dom[Domains] --> Coll[Collections]
	Coll --> Ent[Entities]
	Ent --> Own[Owners]
	Ent --> Pol[Policies]
	Ent --> Tag[Tags/Glossary]
	Ent --> SLA[SLAs]
Loading
  • Domains: logical boundaries (e.g., Finance, Marketing) to group data and assign accountability
  • Collections: folders/groups to organize related assets
  • Owners: teams or users (primary + secondary)
  • Policies: rules for access, retention, quality
  • Tags/Glossary: shared vocabulary and classification
  • SLAs: expectations for freshness, quality, and response

Ownership and responsibilities

Roles:

  • Data Owner: accountable for a domain/collection; approves changes
  • Data Steward: manages metadata quality and classifications
  • Data Consumer: uses data subject to policies

Assign ownership:

  1. Navigate to an entity (table/dashboard/topic)
  2. Click Edit → Owners → select team/user
  3. Save; ownership flows into search facets and lineage views

Domains and collections

Create domains to reflect org structure. Within a domain, create collections for projects or functional groupings.

Best practices:

  • Keep domain count manageable (< 20)
  • Use naming conventions (domain-project-collection)
  • Delegate ownership to domain stewards

Glossary and tags

Use glossaries for business terms and tags for classification.

Steps:

  1. Create a glossary → add terms (definition, owner, reviewers)
  2. Define synonyms and related terms
  3. Apply tags/terms to entities manually or automatically (via classification rules)
graph TD
	G[Glossary Term] -->|applied to| T1[Table]
	G --> V1[View]
	G --> D1[Dashboard]
Loading

Policies and access control

OpenMetadata supports role-based access control (RBAC) and policies.

  • Map IdP groups to OM roles (e.g., Steward, Consumer)
  • Create policies to restrict sensitive tags (e.g., PII) to specific roles
  • Use attribute-based rules where supported

Example policy ideas:

  • Only Stewards can edit descriptions on PII-tagged tables
  • Data Owners can approve term additions in their domain

Data quality and SLAs

Define SLAs for key datasets:

  • Freshness: data updated by 8am UTC daily
  • Completeness: null rate < 0.5%
  • Distribution: values within expected ranges

Implement via tests:

  1. Create tests (row count, null %, min/max) in OpenMetadata or via dbt/Great Expectations
  2. Attach tests to entities and schedules
  3. Configure alerts for failures (Slack/Email)

Stewardship workflow

Common flows:

  • Description updates → steward review
  • Glossary term requests → owner approval
  • Classification changes → security review for PII
sequenceDiagram
	participant U as User
	participant S as Steward
	participant O as Owner
	U->>S: Proposes term/description
	S->>O: Reviews and requests approval
	O-->>S: Approves/Rejects
	S-->>U: Merges change / Requests revision
Loading

Reporting and audits

  • Use search facets to list PII-tagged assets and verify ownership
  • Export metadata for audit snapshots if required
  • Track changes via activity feeds

Getting value quickly

Week 1 checklist:

  • Define 5–10 glossary terms
  • Tag top 20 datasets
  • Assign owners and stewards for 3 domains
  • Add 2–3 quality tests per critical table

Troubleshooting

  • Missing owners: ensure users/teams are synced from SSO/SCIM or defined in OM
  • Inconsistent tags: adopt a governance working group to standardize vocabulary
  • Approval delays: automate notifications and define SLAs

Next: Implement Data Quality tests and set alerting in Monitoring & Maintenance.