DataHub Core
Everything you need to evaluate DataHub Core for data governance: what it is, my review, pricing, features and the best alternatives.
By Ari Bajo - data engineer turned writer.
Updated on November 5, 2026
- Product
- DataHub Core
- Category
- Open-source catalog and lineage
- Pricing
- Free, open source
- Paid version
- DataHub Cloud
- Deployment
- Self-hosted
On this page
What is DataHub Core?
DataHub Core is the open-source DataHub catalog (Apache 2.0): metadata ingestion, search, lineage, a glossary and data products, self-hosted.
DataHub Core review
DataHub Core is the best open-source data governance tool for data engineering teams that want a flexible, self-hosted metadata platform and are comfortable operating it themselves across complex, multi-source data environments.
Strengths
- Free and open source, with one of the largest communities in the category.
- Lineage with OpenLineage support and impact analysis.
- Data contracts and data products in the open-source version.
Caveats
- You run it yourself: ingestion, upgrades and scaling are on you.
- The built-in PII classifier was removed after version 1.6.0.5.
- AI assistants and agents are Cloud-only.
Fun facts
DataHub has 12,954 GitHub stars (November 2026).
DataHub Core is licensed under Apache 2.0.
DataHub was created at LinkedIn.
DataHub Core pricing
DataHub Core is free and open source under Apache 2.0. DataHub Cloud is the paid managed version.
Checked November 5, 2026 on the GitHub repository.
DataHub Core features
DataHub Core appears in my lists of the best data catalog, open-source data governance, data contracts, data lineage, data stewardship and data classification tools. It supports 44 of the 89 tracked data governance features.
DataHub Core data catalog
DataHub Core ingests metadata from 100+ sources, with search, a glossary, domains, data products and structured properties.
Compare: data catalog tools
Data Catalog: Open-source metadata platform with 100+ ingestion sources, search, glossary, domains and data products, self-hosted under Apache 2.0 (docs).
Automated Metadata Harvesting: Python ingestion framework with 100+ connectors (new in 2026: Informatica IDMC, Fabric Data Factory, SAP Datasphere, Cube, Pinecone) extracts schemas, lineage and usage on a schedule (docs).
Search & Discovery: Faceted search across all entity types with filters by platform, domain, owner, tag and glossary term, plus saved Views (docs).
AI-Powered Search: Semantic search over assets and Context Documents ships in Core (v1.4.0, Feb 2026), and the `datahub search` CLI adds semantic search (v1.5.0) (github).
Usage & Popularity Statistics: Dataset Usage & Query History shows top queries, top users and query counts per dataset, available in Core and Cloud (docs).
Custom Fields & Templates: Structured Properties add typed, validated custom fields to any entity, available in Core and Cloud (docs).
Business Glossary: Business glossary with term groups, owners and related-term relationships is available in Core and Cloud; change proposals for terms are Cloud only (docs).
Metadata Propagation: The open-source docs-propagation action pushes column descriptions to downstream and sibling columns via lineage, enabled by default in Core (docs).
Catalog Usage Analytics: Core can emit product analytics events to Mixpanel, Amplitude or Google Analytics, but compliance-form analytics dashboards are Cloud only (docs).
Asset Certification: The Certification feature is DataHub Cloud only (Public Beta, Context add-on); Core has deprecation flags and tags but no certification (docs).
DataHub Core data lineage
DataHub Core shows table, column and end-to-end lineage with impact analysis, and accepts OpenLineage events.
Compare: data lineage tools
Data Lineage: Table- and column-level lineage from SQL parsing, connectors, Airflow, Dagster, Spark and OpenLineage, available in Core (docs).
Table-Level Lineage: Dataset-to-dataset lineage is extracted automatically by connectors and SQL parsing (docs).
Column-Level Lineage: Fine-grained column-level lineage from SQL parsing and supported connectors is available in Core (docs).
End-to-End Lineage: Lineage spans ingestion tools, warehouses, orchestrators and BI tools such as Looker, Tableau, Power BI and Sigma (docs).
Impact Analysis: The lineage explorer and impact analysis view list all downstream dependents of a dataset or column before a change (docs).
OpenLineage Support: DataHub accepts OpenLineage events through a REST endpoint and maps them to its lineage model (docs).
DataHub Core data stewardship
Compare: data stewardship tools
Data Stewardship: Custom ownership types, domains and API-driven compliance forms support stewardship in Core; the full forms UI and proposals are Cloud only (docs).
Governance Campaigns: Compliance Forms assign questions across many assets to owners; in Core they are available via API only (docs).
Metadata Automation: The open-source DataHub Actions framework runs event-triggered actions on metadata changes (e.g. propagation, Slack notifications) (docs).
Critical Data Elements: No dedicated CDE register found; criticality can only be modelled with tags, glossary terms or structured properties (docs).
Governance Scorecards: No questionnaire-based maturity scorecard found in Core docs (docs).
Review & Approval Workflows: Change proposals and approval workflows are documented as DataHub Cloud only (docs).
DataHub Core data compliance
Data Compliance: Audit events API and API-driven compliance forms are available in Core; regulatory mapping is not (docs).
Audit Trail: The audit events search API (/openapi/v1/events/audit/search) records metadata changes and actions, available in Core and Cloud (docs).
Regulatory Framework Mapping: No GDPR/HIPAA/SOX framework mapping found; regulations can only be modelled as glossary terms (docs).
Compliance Reporting: Compliance-form analytics dashboards are DataHub Cloud only; Core has no compliance reporting (docs).
Data Activity Monitoring: DataHub ingests query history for usage stats but does not monitor source access in real time (docs).
ROPA: No GDPR Article 30 processing register in DataHub (docs).
Data Retention Policies: Retention settings cover DataHub's own metadata versions, not retention of source data (docs).
Data Subject Rights: No DSAR or right-to-be-forgotten workflows (docs).
DataHub Core collaboration
Collaboration: Slack notifications via the open-source Actions framework and Context Documents are available in Core; Teams and Ask DataHub bots are Cloud only (docs).
Alert Notifications: The open-source Slack action posts metadata-change notifications to Slack channels (docs).
Discussions & Annotations: Context Documents can be linked to assets, but no threaded discussions on assets were found (docs).
Ticketing Integrations: No Jira or ServiceNow integration found in Core docs (docs).
DataHub Core data marketplace
Data Products: Data Products are first-class entities in Core, with multiple products per asset (v1.5.0) and nested products (v1.8.0) (docs).
Data Mesh Support: Domains (with nested domains and domain-scoped policies) and data products support federated ownership in Core (docs).
Data Marketplace: The Data Product Marketplace browse page and access workflows are DataHub Cloud features (Cloud v2.2.0) (docs).
Self-Service Access Requests: Data Access Workflows are DataHub Cloud only (Private Beta) (docs).
Subscription Management: Subscriptions and notifications are marked DataHub Cloud only (docs).
DataHub Core data quality
Compare: data contracts tools
Data Quality: Core stores and displays assertions and incidents, including Snowflake DMF assertions and external results (dbt, Great Expectations); native monitors are Cloud only (docs).
SQL Table Tests: The open assertions spec compiles freshness, volume, column and custom SQL assertions to Snowflake DMFs from the CLI (docs).
Data Profiling: SQL sources can compute table and column profiles (row counts, null counts, min/max, distinct values) during ingestion (docs).
Data Contracts: Data contracts bundling schema, freshness and quality assertions can be created via the GraphQL API; ODCS ingestion added in v1.7.0 (docs).
Data Health Scores: The Data Health dashboard is marked DataHub Cloud only (docs).
Anomaly Detection: AI anomaly detection is marked DataHub Cloud only (docs).
DataHub Core AI governance
AI Governance: MLModel, MLModelGroup and MLFeature entities catalog models with owners and training data alongside datasets (docs).
AI Asset Inventory: MLModel entities carry versions, owners, hyperparameters and metrics; the Agent Registry for AI agents is Cloud only (docs).
AI Lineage: MLModel aspects link models to their training and evaluation datasets in the lineage graph (docs).
AI Agent Access Governance: The self-hosted MCP server acts with a token's privileges, but query-time Search Access Controls are DataHub Cloud only (docs).
AI Regulatory Compliance: No EU AI Act, NIST AI RMF or ISO 42001 templates; model metadata only (docs).
AI Model Health Score: No composite model trust score (docs).
DataHub Core workflows
Workflows: GraphQL, OpenAPI, Python and Java SDKs, the Actions framework and a self-hosted MCP server are available in Core (docs).
REST / GraphQL API: GraphQL API plus OpenAPI/REST endpoints for reading and writing metadata (docs).
MCP Server: A self-hosted MCP server is documented for DataHub Core; managed and scoped MCP servers are Cloud only (docs).
dbt Integration: The dbt source ingests models, sources, tests, docs and column-level lineage from dbt Core and dbt Cloud (docs).
Python SDK: pip-installable acryl-datahub package with emitters and a high-level Python SDK (pypi).
Webhooks: No built-in outbound webhook; custom actions can be written with the Actions framework (docs).
Browser Extension: The Chrome extension is documented under DataHub Cloud only (docs).
DataHub Core data context layer
Data Context Layer: Metadata graph, Context Documents, Metrics and Semantic Models (beta) and the MCP server are available in Core; ontology/SPARQL is Cloud only (docs).
Knowledge Graph: DataHub's core architecture is a typed metadata graph of entities and relationships queryable via GraphQL (docs).
Semantic Models: Metrics and Semantic Models are first-class entities in Core (Beta, v1.7.0) with lineage through metric upstreams; they are cataloged and linked rather than auto-generated (docs).
Not covered: Data classification · AI features · Master data management
DataHub Core deployment options
DataHub Core is self-hosted, on Kubernetes or Docker.
Deployment: Self-hosted via Docker or Kubernetes Helm charts on AWS, GCP or Azure (docs).
Self-Hosted Deployment: Kubernetes Helm charts and deployment guides for AWS, GCP and Azure (docs).
SaaS Deployment: Core is self-hosted only; the managed SaaS is DataHub Cloud (website).
Hybrid Deployment: The in-VPC remote executor for a cloud control plane is a DataHub Cloud feature (docs).
Data Residency: Not applicable: no vendor-hosted SaaS; residency is wherever you deploy it (docs).
DataHub Core security
DataHub Core supports SSO and role-based access.
Security: OIDC SSO, metadata access policies and personal access tokens are available in Core; SCIM and query-time search ACLs are Cloud only (docs).
SSO: OIDC SSO with providers such as Okta, Azure AD and Google can be configured on self-hosted DataHub (docs).
RBAC: Roles and fine-grained platform and metadata policies, including domain-scoped and glossary-term-targeted policies (v1.5.0) (docs).
Audit Log: The audit events search API logs user and system actions in Core (docs).
Compliance Certifications: Self-hosted open source carries no vendor attestation; SOC 2 applies to DataHub Cloud only (website).
SCIM Provisioning: SCIM provisioning is documented only for DataHub Cloud (Okta) (docs).
DataHub Core alternatives
The closest DataHub Core alternatives and most popular competitors are OpenMetadata, Unity Catalog OSS, Apache Atlas, Collate and Atlan. They share DataHub Core's main use cases and features, from the closest match down, and each one says when to consider it instead.
Choose DataHub Core for a free, self-hosted catalog with strong lineage. For the same with AI agents and no operations, consider DataHub Cloud or Collate.
OpenMetadata
OpenMetadata is an open-source context platform that unifies metadata for humans and AI agents, with data discovery, data quality tests, column-level lineage, and governance.
The closest open-source rival, with data quality tests built in.
Unity Catalog OSS
Unity Catalog OSS is an open-source universal catalog for data and AI (LF AI & Data) for tables, unstructured data, and AI assets, with access control and Iceberg REST APIs.
Consider Unity Catalog OSS if you mainly need an open lakehouse catalog with access control.
Apache Atlas
Apache Atlas is an open-source data catalog with data lineage, a business glossary, data classification, tag-based data policy enforcement, and data governance for Hadoop ecosystems.
Consider Apache Atlas only for Hadoop-centric estates that need policy-driven classification.
Collate
Collate is a managed agentic data platform built on OpenMetadata, with AI agents, data quality tests, observability metrics, column-level lineage, and governance workflows.
Consider Collate for a managed catalog with AI agents and no operations, on an open-source core.
Atlan
Atlan is a context layer for AI, built on a data catalog — with AI-generated asset documentation, column-level data lineage, a business glossary, AI governance, and an MCP server that feeds business context to AI tools.
Consider Atlan for a managed catalog built around AI agents, if you don't need open source.
DataHub Core FAQ
Is DataHub Core open source?
Yes. Apache 2.0 licensed, about 13k GitHub stars (2026-11-05).
How much does DataHub Core cost?
DataHub Core is free and open source under Apache 2.0. DataHub Cloud is the paid managed version.
Can DataHub Core be deployed on-premise?
Yes. DataHub Core is self-hosted, on Kubernetes or Docker.
What are the best DataHub Core alternatives and competitors?
The closest alternatives are OpenMetadata, Unity Catalog OSS, Apache Atlas, Collate and Atlan. The Alternatives section above says when to consider each one over DataHub Core.
Browse all 94 tool profiles, grouped by category.
Evaluating data tools in 2026?
I curate the most comprehensive lists of data tools, up-to-date comparison tables, and actionable market guides for data teams. More useful than a Gartner report. More reliable than asking AI. Next up: Data FinOps, MLOps, Data Orchestration...