--- title: "DataHub Alternatives: 9 Options Compared for 2026" url: "https://atlan.com/know/ai-agent/context-layer/datahub-alternatives/" description: "DataHub alternatives on four paths: DataHub Cloud, open source, managed catalogs and platform-native options, compared on license, coverage and MCP support." author: "Emily Winks" author_role: "Data Governance Expert" published: "2026-11-01" updated: "2026-11-01" --- ## What are the main DataHub alternatives? DataHub alternatives fall into four paths, set by what your team needs. Move to DataHub Cloud for a managed deployment. Stay open source with OpenMetadata or Unity Catalog OSS. Buy a managed catalog that spans many platforms: Alation, Atlan or Collibra. Or consolidate on one platform's native catalog: Databricks Unity Catalog, Google Knowledge Catalog or Microsoft Purview Unified Catalog. Score each finalist on four criteria from its own docs (who runs it, license, cross-platform coverage and AI-agent delivery), then pilot one domain with a lineage diff. The four paths off DataHub: - **Keep DataHub, managed:** DataHub Cloud, the managed edition built on DataHub Core. - **Stay open source:** OpenMetadata (or Collate) and Unity Catalog OSS, both Apache 2.0. - **Managed, many platforms:** Alation, Atlan and Collibra, commercial and vendor-run. - **Consolidate on one platform:** Databricks Unity Catalog, Google Knowledge Catalog or Purview Unified Catalog. ## Key takeaways - DataHub Core is Apache 2.0; its hosted MCP server, fine-grained access control and 99.5% SLA are listed as Cloud-only. - Nine options sort into four paths: DataHub Cloud, open source, managed multi-platform, or one platform's native catalog. - Seven of the nine document an MCP server, so compare release status and read or write scope, not the logo. - Migrate by exporting curated glossary terms and owners, mapping URNs, then diffing lineage on one pilot domain. --- DataHub alternatives sort into four paths, set by what your team needs: DataHub Cloud, another open-source project, a managed catalog that spans many platforms (Alation, Atlan, Collibra), or one platform's native catalog. The 9 options below are compared on 4 criteria from each vendor's own docs, as of November 1, 2026. ### Build Your DataHub Alternatives Shortlist Give it the options you are weighing and every system in your stack. It returns a scored evaluation matrix and the two questions that will decide it. [Read the skill](/skills/catalog-tool-shortlist.md). *Paste into a new chat* ``` Use the skill at https://atlan.com/skills/catalog-tool-shortlist.md to evaluate DataHub alternatives against our stack. Ask me for whatever it needs. ``` *Run once in a terminal* ``` curl -fsSL --create-dirs \ -o ~/.agents/skills/catalog-tool-shortlist/SKILL.md \ https://atlan.com/skills/catalog-tool-shortlist.md ``` *For an agent* ``` curl -fsSL https://atlan.com/skills/catalog-tool-shortlist.md ``` Human Agent ClaudeChatGPTGeminiCursorOther Copy DataHub is an active Apache 2.0 project that calls itself ["The Context Platform for AI Agents" (DataHub, 2026)](https://datahub.com/), with [140+ source connectors in both editions (DataHub docs, 2026)](https://docs.datahub.com/docs/managed-datahub/managed-datahub-overview). For background, see [what LinkedIn's DataHub is and how it works](https://atlan.com/linkedin-datahub-metadata-management-open-source/). Each vendor states coverage in its own unit, so the figures do not compare one to one. | Option | Decision path | Who runs it | License | Coverage, as the vendor states it | AI-agent delivery (vendor docs) | |---|---|---|---|---|---| | DataHub Cloud | 1. Keep DataHub, managed | DataHub, fully managed; in-VPC remote ingestion agent | Commercial, built on Apache 2.0 DataHub Core | 140+ source connectors | Hosted MCP server; scoped MCP servers (private beta); Ask DataHub AI agent | | OpenMetadata (self-host or Collate) | 2. Stay open source | You, or Collate (multi-tenant SaaS on Free; single-tenant, hybrid or private BYOC on paid tiers) | Apache 2.0 | 130+ connectors | MCP server at `/mcp`, enabled by default | | Unity Catalog OSS | 2. Stay open source | You | Apache 2.0, LF AI & Data sandbox project | Delta Lake, Iceberg and Hudi via UniForm; Parquet, JSON, CSV | Functions and AI models as catalog assets | | Alation | 3. Managed, many platforms | Alation (Alation Cloud Service) or you (customer-managed) | Commercial; AI Agent SDK Apache 2.0 | 120+ connectors | Local and remote MCP server through the AI Agent SDK | | Atlan | 3. Managed, many platforms | Atlan, single-tenant SaaS; Context Lakehouse storage can sit in your S3, GCS or ADLS | Commercial | 110 connectors in 15 categories | Hosted MCP server; A2A, SQL, REST and Graph APIs | | Collibra | 3. Managed, many platforms | Collibra Cloud sites run by Collibra, or Edge sites in your environment | Commercial | 100+ native connectors | Hosted Collibra MCP server with read and write tools | | Databricks Unity Catalog | 4. Consolidate on one platform | Databricks | Commercial | Lakehouse Federation: 11 query-federation and 7 catalog-federation sources | Governs MCP services as catalog objects; Genie One MCP GA | | Google Knowledge Catalog | 4. Consolidate on one platform | Google Cloud | Commercial | dbt metadata import GA; Oracle, MySQL, SQL Server, PostgreSQL connectors in Preview | Knowledge Catalog remote MCP server; lineage MCP server (Preview) | | Microsoft Purview Unified Catalog | 4. Consolidate on one platform | Microsoft (SaaS); Data Map scans can use a self-hosted integration runtime | Commercial | Data Map sources across Azure, databases, files, and services and apps | AI-powered copilot search; Unified Catalog REST API (Public Preview) | --- ## Why do teams look for DataHub alternatives? Four situations lead teams to compare options: they want someone else to run the infrastructure, they need a feature DataHub lists for its Cloud edition, the catalog's audience is widening, or they are consolidating on one platform. Each maps to a path. ### What does it take to self-host DataHub Core? [DataHub's Kubernetes deployment guide (DataHub docs, 2026)](https://docs.datahub.com/docs/deploy/kubernetes) lists four dependencies: Kafka, a relational database (MySQL, Postgres or MariaDB), an Elasticsearch search index, and a graph index on Neo4j or Elasticsearch. For a local Minikube setup it notes that "more than 7GB of RAM is required to run DataHub and its dependencies." DataHub's [Breaking Changes by Version doc (DataHub, 2026)](https://github.com/datahub-project/datahub/blob/master/docs/how/breaking-changes.md) asks you to read the entries for every version between yours and the target. Its entry for the next release after v1.8.0 sets the search backend to Elasticsearch 8.x or OpenSearch 2.x or 3.x. Compare running costs with the [build-versus-buy criteria for metadata tooling](https://atlan.com/know/ai-agent/context-layer/metadata-tooling-build-vs-buy-evaluation-criteria/), [the cost of a DIY context layer](https://atlan.com/know/ai-agent/context-layer/diy-context-layer/) and [total cost of ownership across build, buy and bundle](https://atlan.com/know/ai-agent/context-layer/context-layer-tco-build-vs-buy-vs-bundle/). ### Which DataHub features are listed for Cloud only? These rows come from DataHub's Core vs Cloud table in the [DataHub Cloud overview (DataHub docs, 2026)](https://docs.datahub.com/docs/managed-datahub/managed-datahub-overview). | Capability | DataHub Core | DataHub Cloud | |---|---|---| | 140+ connectors, column-level lineage, glossary, data contracts, GraphQL and MCP retrieval, full API and SDK | Yes | Yes | | DataHub hosted MCP server and scoped MCP servers | No | Yes | | AI documentation generation and automated context updates | No | Yes | | Change proposals and access request workflows | No | Yes | | Fine-grained access control and AWS PrivateLink support | No | Yes | | 99.5% uptime SLA and fully managed cloud deployment | No | Yes | Both editions cover lineage, which [DataHub's OpenLineage support](https://atlan.com/know/ai-agent/data-for-ai/datahub-openlineage-support/) extends, and [data contracts](https://atlan.com/datahub-data-contracts/). The Cloud rows serve stewards, analysts and agents; see [role-based access control in a context layer](https://atlan.com/know/ai-agent/context-layer/context-layer-role-based-access-control/), the [agentic data catalog](https://atlan.com/know/agentic-data-catalog/) and the [difference between a data catalog and a context layer](https://atlan.com/know/data-catalog-vs-context-layer/). --- ## How should you evaluate a DataHub alternative? Score every option on four criteria from its own docs. The list is ordered by decision path, 1 to 4, then alphabetically. Two checkable facts set the path: who runs the tool, and whether it is tied to one data platform. | Criterion | What to check in the vendor's own docs | |---|---| | Who runs it | Managed SaaS, single-tenant, BYOC or self-hosted; what runs in your network | | License | Apache 2.0 (as DataHub Core is) or commercial | | Cross-platform coverage | Connector figure and federation scope, against the sources you run | | AI-agent delivery | MCP server, GA or Preview, read or write | Use the [context layer evaluation criteria](https://atlan.com/know/ai-agent/context-layer/context-layer-evaluation-criteria/) and a [data catalog RFP template](https://atlan.com/know/data-catalog/data-catalog-rfp-template-with-evaluation-criteria/), then test finalists against your [context layer reference architecture](https://atlan.com/know/ai-agent/context-layer/context-layer-reference-architecture/). Agent protocols are still being defined, which is why the AI-agent criterion asks for release status. Mastercard CDO Andrew Reiskind discusses them in this Re:Govern 2025 session. --- ## Path 1: Should you stay on DataHub and move to DataHub Cloud? DataHub describes its managed edition as ["The enterprise context platform built on DataHub Core" (DataHub, 2026)](https://datahub.com/). **Pros** - Fully managed cloud deployment with a 99.5% uptime SLA and an in-VPC remote ingestion agent, per the [DataHub Cloud overview (DataHub docs, 2026)](https://docs.datahub.com/docs/managed-datahub/managed-datahub-overview). - A hosted MCP server with read tools such as `search`, `get_lineage` and `draft_sql_for_tables`, and mutation tools on Cloud v0.3.17+, per [DataHub's MCP guide (DataHub docs, 2026)](https://docs.datahub.com/docs/features/feature-guides/mcp). **Good to know** - The hosted MCP server needs DataHub Cloud v0.3.12+, and scoped MCP servers are labeled "Context Platform private beta" in the same guide. - Acryl Data, Inc. is the company named in the datahub.com footer. **Choose DataHub Cloud if** DataHub fits your team and you want a managed deployment. --- ## Path 2: Which open-source DataHub alternatives are active? OpenMetadata and Unity Catalog OSS are active Apache 2.0 projects. Amundsen's [repository (GitHub, 2026)](https://github.com/amundsen-io/amundsen) "was archived by the owner on Sep 29, 2026"; see [Amundsen alternatives](https://atlan.com/amundsen-alternatives/). ### OpenMetadata and Collate OpenMetadata describes itself as ["The Open Context Layer for Data and AI" (OpenMetadata, GitHub, 2026)](https://github.com/open-metadata/OpenMetadata). It is Apache 2.0, lists 130+ connectors, and exposes an MCP server at `/mcp` for searching metadata, inspecting lineage and updating descriptions, tags and owners. **Pros** - The MCP server "ships installed and enabled by default," with OAuth 2.0 or token authentication, per [OpenMetadata's MCP guide (OpenMetadata docs, 2026)](https://docs.open-metadata.org/latest/how-to-guides/mcp). - [Collate's pricing page (Collate, 2026)](https://www.getcollate.io/pricing) lists a Free tier with 5 users and 500 data assets on multi-tenant SaaS. **Good to know** - Collate Premium includes 25 users and 5,000 assets on single-tenant or hybrid SaaS; Enterprise includes 50 users and 10,000 assets and adds private BYOC. **Choose OpenMetadata if** open source is a requirement. See [what OpenMetadata is used for](https://atlan.com/know/ai-agent/data-for-ai/what-is-openmetadata-used-for/) and [OpenMetadata vs DataHub](https://atlan.com/openmetadata-vs-datahub/). ### Unity Catalog OSS Unity Catalog OSS is the ["Open, Multi-modal Catalog for Data & AI" (Unity Catalog, GitHub, 2026)](https://github.com/unitycatalog/unitycatalog), Apache 2.0 licensed. **Pros** - Supports Delta Lake, Apache Iceberg and Apache Hudi via UniForm, plus Parquet, JSON and CSV, and catalogs tables, files, functions and AI models. - Release 0.7.0 (October 13) added Iceberg table support through an Iceberg REST Catalog, per the [Unity Catalog releases page (GitHub, 2026)](https://github.com/unitycatalog/unitycatalog/releases). **Good to know** - The README states it "is currently a sandbox project with LF AI and Data Foundation." **Choose Unity Catalog OSS if** your estate is lakehouse-first and you want to run the catalog. --- ## Path 3: Which managed DataHub alternatives work across many platforms? Alation, Atlan and Collibra are commercial, vendor-run catalogs that connect to many platforms, listed alphabetically. See [a multi-cloud context layer](https://atlan.com/know/ai-agent/context-layer/multi-cloud-context-layer/) and [what an AI context platform is](https://atlan.com/know/ai-agent/context-layer/ai-context-platform/). ### Alation Alation's data catalog page (alation.com, accessed 2026-11-01) lists 120+ pre-built connectors and Alation Anywhere, which brings the catalog into Excel, Teams and Slack. **Pros** - ALLIE AI recommends metadata descriptions during curation, per the same page. - Alation Cloud Service is "fully administered, maintained, and upgraded by Alation"; a customer-managed option also exists, per Alation's Cloud Service vs Customer-Managed doc (accessed 2026-11-01). - The [Alation AI Agent SDK (GitHub, 2026)](https://github.com/Alation/alation-ai-agent-sdk) is Apache 2.0, with a local MCP server and a remote one at `https:///ai/mcp`. **Good to know** - The SDK's October 23, 2026 update calls 1.0.0 "the first stable release." - Its January 2, 2026 update moved `update_catalog_asset_metadata` and `check_job_status` work to the Alation REST API. **Choose Alation if** business-user adoption in Office and chat tools comes first. ### Atlan Atlan's [connectors page (Atlan, 2026)](https://atlan.com/connectors/) lists 110 connectors in 15 categories. Its [Context Lakehouse (Atlan, 2026)](https://atlan.com/context-lakehouse/) stores context on "Apache Iceberg underneath," which "can live in your own cloud storage (S3, GCS, ADLS)," and serves it through MCP, A2A, SQL, and REST and Graph APIs. **Pros** - A hosted remote MCP server with per-user OAuth or an API key, per the [Atlan remote MCP overview (Atlan docs, 2026)](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview). More in [Atlan's MCP server](https://atlan.com/know/what-is-atlan-mcp/). - [Context Agents (Atlan, 2026)](https://atlan.com/context-agents/) draft descriptions, glossary terms and domain tags from query history, lineage and SQL usage, the work [context mining](https://atlan.com/know/ai-agent/context-engineering/context-mining-for-ai/) describes. **Good to know** - [Atlan's architecture doc (Atlan docs, 2026)](https://docs.atlan.com/platform/references/atlan-architecture) describes "a cloud-first, single-tenant SaaS platform" on AWS, Azure or GCP; the Self-Deployed Runtime for connectors runs in your network. - Its 110 listed connectors compare with 140+ for DataHub and 130+ for OpenMetadata. - The MCP server "exposes both read and write tools" by default; metadata writes follow the "Create and update metadata from conversational AI" Labs setting, and read-only mode is available on request, per [Atlan's MCP security doc (Atlan docs, 2026)](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security). **Choose Atlan if** you want context in open storage you control, served to agents through several interfaces. See [Atlan vs DataHub](https://atlan.com/atlan-vs-datahub/). ### Collibra Collibra's data catalog page (collibra.com, accessed 2026-11-01) lists 100+ native connectors, Collibra AI Copilot, data contracts, and publishing data products to the Collibra Data Marketplace. **Pros** - A hosted MCP server with per-user OAuth 2.0, per Collibra's MCP integration doc (dated 2026-08-20). Collibra's MCP server page calls it "a production-ready Model Context Protocol server" that "supports both read and write operations." - Write tools include `create_asset` and `edit_asset`, per Collibra's MCP tools reference (dated 2026-10-28). **Good to know** - Collibra Cloud sites "are managed and deployed by Collibra," or an Edge site runs in your environment, per Collibra's About Edge doc (accessed 2026-11-01). - Per the tools reference, "Data quality tools are generally available, but they are off by default." **Choose Collibra if** governance policy and contracts lead the program. --- ## Path 4: Which platform-native catalogs replace DataHub when you consolidate? If one cloud or lakehouse runs most of your estate, its native catalog can take over DataHub's role. Check federation scope beyond that platform; [single-stack lock-in versus a neutral context layer](https://atlan.com/know/ai-agent/context-layer/single-stack-lock-in-vs-neutral-context-layer/) covers the trade-off. ### Databricks Unity Catalog Databricks' "What is Unity Catalog?" doc (docs.databricks.com, updated 2026-09-11) calls it "the unified governance layer for data and AI built into Databricks," governing "tables, views, volumes, functions, models, and services (model services and MCP services)." **Pros** - The Lakehouse Federation doc (updated 2026-09-29) lists 11 query-federation sources, including Snowflake and BigQuery, and 7 catalog-federation sources, including AWS Glue. - "The Genie One MCP is generally available," per Databricks' managed MCP doc (updated 2026-10-23); each `system.ai` MCP is "a Unity Catalog securable invoked through Unity Gateway." **Good to know** - Catalog federation "Runs directly against object storage on Databricks compute only," per the federation doc. - The `dbsql`, `sandbox` and `web_search` MCPs need the Unity Gateway beta, and workspace MCP endpoints "are in Public Preview." **Choose Unity Catalog if** Databricks is the center of gravity. See [a context layer for Databricks](https://atlan.com/know/context-layer-for-databricks/) and the [MCP server for Databricks](https://atlan.com/know/mcp/mcp-server-for-databricks/). ### Google Knowledge Catalog Knowledge Catalog is the April 10, 2026 rename of Dataplex Universal Catalog; API, client library, CLI and IAM names stayed the same, per the [Knowledge Catalog release notes (Google Cloud, 2026)](https://docs.cloud.google.com/dataplex/docs/release-notes). **Pros** - dbt metadata import reached GA on September 24, 2026, and Looker (Google Cloud core) lineage on October 26, 2026. - The remote MCP server "lets you discover data assets, search metadata, retrieve entry details, and manage data products," with read-only or read-write OAuth scopes, per [Google's remote MCP guide (Google Cloud, updated 2026-10-29)](https://docs.cloud.google.com/dataplex/docs/use-remote-mcp). The [Knowledge Catalog AI overview (Google Cloud, updated 2026-10-29)](https://docs.cloud.google.com/dataplex/docs/ai-overview) adds a local MCP Toolbox and a LookupContext API that returns "a unified YAML or JSON context payload for data assets." **Good to know** - Data products MCP support, the lineage remote MCP server, and the Oracle, MySQL, SQL Server and PostgreSQL connectors are Preview in the release notes. **Choose Knowledge Catalog if** BigQuery runs most of your analytics. See [Google Knowledge Catalog](https://atlan.com/know/ai-agent/gcp/google-knowledge-catalog/). ### Microsoft Purview Unified Catalog [Learn about Microsoft Purview Unified Catalog (Microsoft Learn, updated 2026-03-16)](https://learn.microsoft.com/en-us/purview/unified-catalog) describes a catalog organized by governance domains, with data products, critical data elements, glossary terms that carry policies, data quality, health controls and OKRs. **Pros** - Search by domain, data product or keyword, or through "the AI powered copilot," with access requests made in the catalog, per the [same Unified Catalog page (Microsoft Learn, 2026)](https://learn.microsoft.com/en-us/purview/unified-catalog). - [Purview Data Map's source list (Microsoft Learn, updated 2026-07-01)](https://learn.microsoft.com/en-us/purview/data-map-data-sources) covers Azure services plus Snowflake, Google BigQuery, Teradata, Oracle, Fabric, Power BI, Looker, Tableau and Salesforce. **Good to know** - Access requires the enterprise version of the new Microsoft Purview experience, which is rolling out by region. - The [Unified Catalog API (Microsoft Learn, 2026)](https://learn.microsoft.com/en-us/rest/api/purview/unified-catalog-api-overview) is in Public Preview (latest version `2026-03-20-preview`) and covers Unified Catalog features that are GA. **Choose Purview Unified Catalog if** Microsoft runs your estate. See [Purview Unified Catalog](https://atlan.com/know/ai-agent/microsoft/purview-unified-catalog/) and whether to [coexist with, extend or replace Purview](https://atlan.com/know/ai-agent/microsoft/coexist-extend-replace-purview/). --- ## How do DataHub and its alternatives compare on AI-agent delivery? Seven of the nine options document a server for the Model Context Protocol, "an open-source standard for connecting AI applications to external systems," per the [MCP docs (Model Context Protocol, 2026)](https://modelcontextprotocol.io/docs/getting-started/intro). Compare release status and read or write scope; each profile above lists both, as of November 1, 2026. DataHub Core users can run the Apache 2.0 [mcp-server-datahub (GitHub, 2026)](https://github.com/acryldata/mcp-server-datahub), with read tools by default and mutation tools opt-in via `TOOLS_IS_MUTATION_ENABLED=true`. For how agents use these servers, see an [MCP-connected data catalog](https://atlan.com/know/mcp-connected-data-catalog/) and [MCP for data lineage](https://atlan.com/know/mcp/mcp-for-data-lineage/). --- ## How do you migrate off DataHub without losing glossary terms and owners? Export what people curated, map identities, re-create it in the target, then check lineage on one domain. DataHub's docs give the mechanics. 1. **Inventory and export.** List the glossary terms, tags, domains, owners, documentation and contracts people use. [DataHub's DataHub source (DataHub docs, 2026)](https://docs.datahub.com/docs/generated/ingestion/sources/datahub) exists to "Migrate data from one DataHub instance to another," needs access to the instance's database, Kafka broker and schema registry, and can write to the [file sink (DataHub docs, 2026)](https://docs.datahub.com/docs/metadata-ingestion/sink_docs/metadata-file) instead of another instance. 2. **Map identities and glossary structure.** DataHub identifies resources by [URN (DataHub docs, 2026)](https://docs.datahub.com/docs/what/urn), such as `urn:li:corpuser:jdoe` for a user; map each owner to an identity in the target's SSO. The [business glossary source (DataHub docs, 2026)](https://docs.datahub.com/docs/generated/ingestion/sources/business-glossary) lists term fields including owners, `inherits`, `contains` and domain, so check the target keeps that hierarchy. Settle [who owns context, data or AI teams](https://atlan.com/know/context-layer-ownership-data-vs-ai-teams/), and plan for [managing KPI definitions and a business glossary](https://atlan.com/know/ai-agent/how-to-manage-kpi-definitions-and-a-business-glossary-in-a-data-catalog/). 3. **Decide replace or coexist.** To coexist, connect both tools to the same sources and pick one as the system of record for glossary terms and owners. 4. **Pilot one domain and diff lineage.** For the tables agents query most, pull edges with DataHub's [lineage API (DataHub docs, 2026)](https://docs.datahub.com/docs/api/tutorials/lineage), which returns one hop by default and the full graph when `max_hops` is above 2, then diff them against the target's own lineage. Run it as a [context layer pilot or proof of value](https://atlan.com/know/ai-agent/context-layer/how-to-start-a-context-layer-pilot-or-proof-of-value/), with [automated SQL lineage](https://atlan.com/know/ai-agent/data-for-ai/automated-sql-lineage-vs-manual-lineage-mapping/) on the target side. --- ## When should you stay with DataHub instead of switching? Stay with DataHub when your team runs it comfortably, lineage is the main job, and most users are engineers, the audience a [context layer for data engineering teams](https://atlan.com/know/context-layer-for-data-engineering-teams/) serves. If you want the same product with managed operations, start with DataHub Cloud. Otherwise, name your situation, score finalists on the four criteria from their own docs, and pilot one domain with a lineage diff. --- ## FAQs about DataHub alternatives ### 1. What is the difference between DataHub Core and DataHub Cloud? DataHub Core is the Apache 2.0 open-source edition you host yourself. DataHub Cloud is the commercial, fully managed edition built on Core. Cloud adds a hosted MCP server, scoped MCP servers in private beta, AI documentation generation, access request workflows, fine-grained access control and a 99.5% uptime SLA. ### 2. Is OpenMetadata better than DataHub? Neither is better in general. Both are Apache 2.0, list more than 100 connectors and ship an MCP server; the choice is which project your team would rather run. ### 3. Can DataHub serve context to AI agents through MCP? Yes. The open-source MCP server has read tools such as search and lineage, plus opt-in mutation tools. DataHub Cloud adds a hosted server (v0.3.12+, mutation tools on v0.3.17+), scoped servers in private beta and the Ask DataHub AI agent. ### 4. Is Acryl Data the same as DataHub? Acryl Data, Inc. is the company named in the datahub.com footer. The product is DataHub, in Core and Cloud editions. ### 5. Can you keep DataHub and add another tool on top of it? Yes. Connect both tools to the same warehouses, BI tools and pipelines, and pick one as the system of record for glossary terms and owners. ### 6. What should you carry over when you move off DataHub? Glossary terms, tags, domains, owners, documentation and data contracts. Export them with DataHub's DataHub source and a file sink, map owner URNs to the target's identities, and diff lineage on one pilot domain before cut-over. --- ## Sources 1. [DataHub Cloud overview and Core vs Cloud features, DataHub](https://docs.datahub.com/docs/managed-datahub/managed-datahub-overview), accessed 2026-11-01 2. [Deploying DataHub with Kubernetes, DataHub](https://docs.datahub.com/docs/deploy/kubernetes), accessed 2026-11-01 3. [Breaking Changes by Version, DataHub (GitHub)](https://github.com/datahub-project/datahub/blob/master/docs/how/breaking-changes.md), "Unreleased" entry after v1.8.0, accessed 2026-11-01 4. [DataHub homepage, DataHub](https://datahub.com/), accessed 2026-11-01 5. [MCP Server feature guide, DataHub](https://docs.datahub.com/docs/features/feature-guides/mcp), accessed 2026-11-01 6. [mcp-server-datahub, Acryl Data (GitHub)](https://github.com/acryldata/mcp-server-datahub), accessed 2026-11-01 7. [DataHub source (GA), DataHub](https://docs.datahub.com/docs/generated/ingestion/sources/datahub), accessed 2026-11-01 8. [File sink, DataHub](https://docs.datahub.com/docs/metadata-ingestion/sink_docs/metadata-file), accessed 2026-11-01 9. [Business Glossary source, DataHub](https://docs.datahub.com/docs/generated/ingestion/sources/business-glossary), accessed 2026-11-01 10. [URNs, DataHub](https://docs.datahub.com/docs/what/urn), accessed 2026-11-01 11. [Lineage API tutorial, DataHub](https://docs.datahub.com/docs/api/tutorials/lineage), accessed 2026-11-01 12. [OpenMetadata, OpenMetadata (GitHub)](https://github.com/open-metadata/OpenMetadata), accessed 2026-11-01 13. [MCP how-to guide, OpenMetadata docs](https://docs.open-metadata.org/latest/how-to-guides/mcp), docs v2.0.x, accessed 2026-11-01 14. [Collate pricing, Collate](https://www.getcollate.io/pricing), accessed 2026-11-01 15. [Unity Catalog OSS, Unity Catalog (GitHub)](https://github.com/unitycatalog/unitycatalog), accessed 2026-11-01 16. [Unity Catalog releases, Unity Catalog (GitHub)](https://github.com/unitycatalog/unitycatalog/releases), release 0.7.0, accessed 2026-11-01 17. [Amundsen repository, archived 2026-09-29, Amundsen (GitHub)](https://github.com/amundsen-io/amundsen), accessed 2026-11-01 18. Alation Data Catalog, Alation, alation.com/product/data-catalog/, accessed 2026-11-01 19. Alation Cloud Service vs. Customer-Managed Alation, Alation, docs.alation.com/en/latest/welcome/CloudAndOnPrem/index.html, accessed 2026-11-01 20. [Alation AI Agent SDK, Alation (GitHub)](https://github.com/Alation/alation-ai-agent-sdk), accessed 2026-11-01 21. [Atlan connectors, Atlan](https://atlan.com/connectors/), accessed 2026-11-01 22. [Context Lakehouse, Atlan](https://atlan.com/context-lakehouse/), accessed 2026-11-01 23. [Context Agents, Atlan](https://atlan.com/context-agents/), accessed 2026-11-01 24. [Remote MCP overview, Atlan docs](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview), accessed 2026-11-01 25. [Remote MCP security, Atlan docs](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security), accessed 2026-11-01 26. [Atlan architecture, Atlan docs](https://docs.atlan.com/platform/references/atlan-architecture), accessed 2026-11-01 27. Collibra Data Catalog, Collibra, collibra.com/products/data-catalog, accessed 2026-11-01 28. Collibra MCP Server, Collibra, collibra.com/products/mcp-server, accessed 2026-11-01 29. Integrating with Collibra's MCP server, Collibra, productresources.collibra.com/docs/collibra/latest/Content/ModelContextProtocol/co_integrating-with-collibra-mcp.htm, page dated 2026-08-20, accessed 2026-11-01 30. MCP tools reference, Collibra, productresources.collibra.com/docs/collibra/latest/Content/ModelContextProtocol/ref_mcp-tools.htm, page dated 2026-10-28, accessed 2026-11-01 31. About Edge, Collibra, productresources.collibra.com/docs/collibra/latest/Content/Edge/co_edge.htm, accessed 2026-11-01 32. What is Unity Catalog?, Databricks, docs.databricks.com/aws/en/data-governance/unity-catalog/, updated 2026-09-11, accessed 2026-11-01 33. Lakehouse Federation, Databricks, docs.databricks.com/aws/en/query-federation/, updated 2026-09-29, accessed 2026-11-01 34. Managed MCP servers, Databricks, docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp, updated 2026-10-23, accessed 2026-11-01 35. [Knowledge Catalog release notes, Google Cloud](https://docs.cloud.google.com/dataplex/docs/release-notes), accessed 2026-11-01 36. [Use the Knowledge Catalog remote MCP server, Google Cloud](https://docs.cloud.google.com/dataplex/docs/use-remote-mcp), updated 2026-10-29, accessed 2026-11-01 37. [Knowledge Catalog AI overview, Google Cloud](https://docs.cloud.google.com/dataplex/docs/ai-overview), updated 2026-10-29, accessed 2026-11-01 38. [Learn about Microsoft Purview Unified Catalog, Microsoft Learn](https://learn.microsoft.com/en-us/purview/unified-catalog), updated 2026-03-16, accessed 2026-11-01 39. [Data sources that connect to Microsoft Purview Data Map, Microsoft Learn](https://learn.microsoft.com/en-us/purview/data-map-data-sources), updated 2026-07-01, accessed 2026-11-01 40. [Unified Catalog API (Public Preview), Microsoft Learn](https://learn.microsoft.com/en-us/rest/api/purview/unified-catalog-api-overview), updated 2026-04-30, accessed 2026-11-01 41. [What is the Model Context Protocol (MCP)?, Model Context Protocol](https://modelcontextprotocol.io/docs/getting-started/intro), accessed 2026-11-01