The closest open source alternative is OpenMetadata, which also has a managed edition, Collate, with a free tier and published plan limits. For Hadoop estates governed with Apache Ranger, Apache Atlas is still maintained. If your stack sits mostly in one cloud or platform, the built-in catalog may remove the need to run one at all: AWS Glue Data Catalog on AWS, Knowledge Catalog (formerly Dataplex Universal Catalog) on Google Cloud, Microsoft Purview on Azure and Fabric, and Unity Catalog on Databricks, which also has an open source server. If you want a commercial SaaS catalog rather than an open source one, Atlan. Amundsen was archived in September 2026 and is not an option.
DataHub is an open source metadata platform under the Apache 2.0 licence for search, lineage and governance across a data stack. DataHub Core is free to self-host on Docker or Kubernetes; DataHub Cloud is the managed edition from the company that commercialises the project, which was called Acryl Data until it rebranded as DataHub in May 2025. Core v1.8.0 shipped on 22 October 2026, with v1.9.0 release candidates in November.
Teams that look at alternatives usually give one of these reasons. The first three follow from DataHub's own pages; the last is our editorial reading.
- Running it yourself. Self-hosting Core means operating several services, including a search engine (OpenSearch or Elasticsearch), and handling upgrades.
- Cloud-only features. Observability monitors, access workflows and the uptime SLA are in DataHub Cloud, not Core.
- No published pricing. datahub.com shows no plans, prices or free managed tier for DataHub Cloud.
- One-cloud stacks (editorial). Teams whose data sits almost entirely in AWS, Google Cloud, Azure or Databricks may already pay for a catalog that covers their main need.
For the direct comparison, see DataHub vs OpenMetadata; for the platform catalogs, see Unity Catalog vs Snowflake Horizon and Microsoft Purview vs Informatica. If you are choosing a commercial catalog, see Alation alternatives (discovery-led) and Collibra alternatives (governance-led).
Quick picks
Apache 2.0 catalog with built-in data quality tests, and a Collate managed tier that is free for 5 users.
Classification-driven tag policies and lineage for Hive, HBase and Kafka, still maintained by the ASF.
One managed metastore for Athena, EMR, Redshift and Glue, free for the first million objects.
Built into Databricks Premium and Enterprise, and an Apache 2.0 server with Iceberg REST support.
Single-tenant SaaS with lineage, glossary, APIs and an MCP server, with nothing to host.
How we chose
We considered OpenMetadata with Collate, Apache Atlas, Amundsen, Unity Catalog (open source and in Databricks), AWS Glue Data Catalog, Google Cloud Knowledge Catalog, Microsoft Purview, Snowflake Horizon Catalog and Atlan. To be included, a product had to be maintained or sold in November 2026 and offer a searchable metadata catalog with lineage or classification, and its licence or pricing model had to be documented on an official page.
Amundsen is not listed: its repository was archived on 29 September 2026 due to inactivity, and its last releases date from 2024. We also wrote one page for this pair rather than a separate OpenMetadata alternatives page, because the open source catalog field is small and that page would list the same products for the same reasons; OpenMetadata users can read this page with DataHub as the first alternative. Snowflake Horizon Catalog is covered on the Alation alternatives page.
For each product we recorded the licence, any managed edition or free tier, deployment model and main limitation. The order is not a ranking: open source catalogs come first, then cloud and platform catalogs, then a commercial SaaS catalog.
At a glance
| Product | Licence or pricing | Managed option | Coverage | Best for | Main trade-off |
|---|---|---|---|---|---|
| OpenMetadata | Apache 2.0; Collate plans by quote | Collate (free tier: 5 users, 500 assets) | Neutral; 130+ connectors | Open source catalog with data quality | Some features Collate-only |
| Apache Atlas | Apache 2.0 | None from the project | Hadoop components (Hive, HBase, Kafka) | Hadoop with Apache Ranger | Source-only releases; no cloud warehouse hooks |
| Unity Catalog | Included in Databricks Premium and Enterprise; OSS server Apache 2.0 | Databricks | Databricks and connected catalogs | Databricks lakehouses | Thin glossary and stewardship |
| AWS Glue Data Catalog | Free first 1M objects and requests, then usage | AWS managed | AWS (Athena, EMR, Redshift, Glue) | S3 and Iceberg lakes on AWS | AWS and region scoped |
| Knowledge Catalog | DCU-hours and metadata storage | Google Cloud managed | BigQuery and Google Cloud assets | BigQuery-centred teams | Limited connectors outside Google Cloud |
| Microsoft Purview | Free version (preview); pay-as-you-go meters | Microsoft managed | Azure, Fabric, Power BI and scanned sources | Microsoft-centred estates | Several meters; support varies by source |
| Atlan | By quote | Single-tenant SaaS | Neutral; 80+ connectors | Teams that want to stop hosting | Proprietary; no self-hosted platform |
OpenMetadata (and Collate)
OpenMetadata is an Apache 2.0 catalog that combines discovery, lineage, a glossary, governance and data quality tests in one platform, with more than 130 connectors. Version 2.0.0 shipped on 24 August 2026, followed by 2.0.x patches to 27 October and a 2.1.0 release candidate on 28 October 2026. Collate, which owns the OpenMetadata trademark, sells the managed service.
The practical differences from DataHub are that data quality tests are part of the open source build and that Collate publishes its plans: a Free Tier (multi-tenant, 5 users, 500 data assets, unlimited connectors), Premium (25 users, 5,000 assets) and Enterprise (50 users, 10,000 assets), the paid plans by sales contact. See DataHub vs OpenMetadata.
- Some features are Collate-only, so check release notes before assuming parity with the open source build.
- The Collate Free Tier is capped at 5 users and 500 data assets.
- Self-hosting still means running the platform and its supporting services yourself.
Apache Atlas
Apache Atlas is the Apache Software Foundation's metadata and governance framework for Hadoop, with typed metadata, lineage, a glossary and classifications that Apache Ranger can use for tag-based access control and masking. It is maintained: 2.5.0 was released on 30 April 2026, and 2.6.0 release candidates were tagged between August and September 2026.
It is a fit when your estate is Hive, HBase and Kafka on Hadoop and access is governed in Ranger; it is not a general catalog for cloud warehouses and BI tools (editorial).
- Source-only releases: you build, package and operate it yourself.
- Documented hooks cover Hadoop components, not cloud warehouses or BI tools.
- No stewardship workflows or managed edition from the project.
Unity Catalog
Unity Catalog is the governance layer built into Databricks, covering tables, volumes, functions, models and AI assets with fine-grained access control, automatic lineage, auditing and Iceberg REST access; Databricks states it needs no separate licence on Premium and Enterprise. Separately, the open source Unity Catalog server (Apache 2.0, an LF AI & Data sandbox project) released 0.7.0 on 13 October 2026 with Iceberg REST create, write, read and drop support.
It replaces DataHub only for part of the job: it is a technical catalog and access layer rather than a discovery portal across every tool (editorial). See Unity Catalog vs Snowflake Horizon.
- Governs Databricks and connected catalogs, not a whole enterprise estate.
- Glossary and stewardship workflows are thinner than in dedicated catalogs.
- The open source edition's README says its APIs are still evolving.
AWS Glue Data Catalog
The Glue Data Catalog is the managed metadata store inside AWS Glue that Athena, EMR, Redshift and Glue jobs share, with crawlers, an Iceberg REST endpoint and fine-grained permissions through AWS Lake Formation. The first million objects stored and the first million requests each month are free; above that AWS lists USD 1.00 per 100,000 objects per month, and crawlers are billed per DPU-hour.
For an AWS-only lake, it already holds the technical metadata that DataHub would ingest, so some teams use it directly and skip a separate portal (editorial). Business features such as a glossary and semantic search are in preview. See our AWS Glue profile.
- AWS and region scoped; not a neutral catalog for other clouds.
- Business glossary and semantic search are only in preview.
- Crawler, statistics and compaction runs add DPU-hour charges.
Knowledge Catalog (Google Cloud)
Knowledge Catalog is Google Cloud's managed metadata, lineage, data quality and governance service. It was called Dataplex Universal Catalog until 10 April 2026, and its API, CLI and IAM names still use Dataplex; the older Data Catalog service began a phased shutdown on 1 June 2026. It indexes BigQuery and other Google Cloud assets automatically and bills processing in Data Compute Units, with premium processing for lineage, quality and profiling, plus metadata storage.
For a BigQuery-centred team it covers discovery, lineage and quality without running anything (editorial). We could not load Google's pricing page on 7 November 2026, so check rates there or in the Google Cloud pricing calculator.
- Google Cloud centric; connectors outside Google Cloud are limited compared with neutral catalogs.
- Repeated renames mean documentation, SKUs and APIs still use Dataplex names.
- Premium processing (lineage, quality, profiling) has no free tier.
Read our Knowledge Catalog (Google Cloud) review · Official site
Microsoft Purview
Purview data governance pairs the Data Map (scanning, classification and lineage) with Unified Catalog (governance domains, data products, glossary and data quality). A free version in preview covers up to 1,000 annotated assets with live view of a few Azure sources and Fabric; the enterprise version bills Unified Catalog per governed asset per day and data quality in processing units.
For Microsoft-centred estates it gives a managed catalog with classification and governance in a portal the organisation already uses (editorial). See Microsoft Purview vs Informatica.
- Several pricing meters, and the pricing page text hides the unit rates.
- Lineage, classification and policy support vary by source.
- The free version is in preview, capped at 1,000 annotated assets, with no support tickets.
Atlan
Atlan is a commercial single-tenant SaaS metadata platform with a catalog, lineage, glossary, governance and data quality signals, more than 80 connectors according to its home page, and APIs, an SDK and an MCP server. Private sources are reached through the Self-Deployed Runtime on Docker Compose or Kubernetes.
It suits a DataHub Core team whose main problem is operating the platform rather than its licence, and who would rather pay for a managed product than run search, storage and ingestion themselves (editorial). Like DataHub Cloud, it publishes no prices.
- No published pricing.
- Proprietary, with no self-hosted option for the platform.
- Teams that used the Secure Agent must move to the Self-Deployed Runtime (deprecated 30 June 2026).
When to stay with DataHub
Stay with DataHub if you already run Core and its main cost, the services and upgrades, is under control: it is actively released, Apache 2.0, has more than 140 connectors, and lets you move to DataHub Cloud later without changing the metadata model (editorial). DataHub Cloud is also the simplest way to get observability monitors and access workflows without switching catalogs.
Switching makes most sense when you want data quality tests and published managed tiers (OpenMetadata and Collate), when your estate is Hadoop with Ranger (Apache Atlas), when one cloud or platform now holds almost everything (Glue Data Catalog, Knowledge Catalog, Purview, Unity Catalog), or when you no longer want to host a catalog at all (Atlan). Export your custom metadata and glossary first; ingestion recipes do not carry over between products.
Frequently asked questions
What is the best open source alternative to DataHub?
OpenMetadata is the closest: also Apache 2.0, with discovery, lineage and governance, plus data quality tests in the open source build. Apache Atlas is maintained but focused on Hadoop. See DataHub vs OpenMetadata.
What are the alternatives to OpenMetadata?
Largely the same products as on this page, with DataHub as the first open source option instead of OpenMetadata. DataHub Cloud is the managed edition, by quote; for Hadoop, Apache Atlas; for single-cloud stacks, the platform catalogs listed above.
Is Amundsen still maintained?
No. The Amundsen repository was archived on 29 September 2026 due to inactivity, and its last releases were in 2024. Teams still running it should plan a move to a maintained catalog.
Is DataHub the same as Acryl Data?
Acryl Data was the company behind DataHub. It rebranded as DataHub in May 2025, and Acryl Cloud became DataHub Cloud. The open source project itself remains Apache 2.0.
Does DataHub Cloud have a free tier?
No free managed tier or price list is published on datahub.com; DataHub Cloud is sold by quote. DataHub Core is free to self-host. Collate publishes a free tier for managed OpenMetadata.
Sources
- DataHub home page
- DataHub Cloud overview
- DataHub repository
- DataHub v1.8.0 release
- DataHub Series B announcement
- OpenMetadata releases
- Collate pricing
- Apache Atlas downloads
- Amundsen repository (archived)
- Databricks Unity Catalog
- Unity Catalog 0.7.0 release
- AWS Glue pricing
- AWS Glue Data Catalog
- Knowledge Catalog release notes
- Microsoft Purview pricing
- Microsoft Purview free version
- Atlan architecture
Checked November 2026.
How we research these guides: our editorial method.