What you will be able to do
- Explain what Databricks Marketplace is for and who publishes to it
- Name the listing types Marketplace offers: datasets (tables and volumes), AI models, notebooks, apps and MCP servers
- Describe how OpenSharing and shares sit underneath every Marketplace listing
- Tell public listings apart from private exchanges, and say what a provider needs to publish
Key concept
Databricks Marketplace — An open exchange where providers publish data, AI and software products. Databricks customers browse, request and connect to those products from their own workspace, and OpenSharing carries the data between the two sides securely.
1.What Marketplace is for
Analysts often need data their organisation doesn't produce, such as reference datasets, commercial feeds or industry benchmarks. Getting it the usual way means one-off contracts, file transfers and loading jobs. Databricks Marketplace replaces that with an open exchange. Data providers, software vendors and technology partners publish offerings there, and Databricks customers discover, evaluate and connect to them directly from their workspace. The documentation describes the result as a single catalog for finding data, software and AI capabilities across the Databricks partner ecosystem.
The exchange has two sides. Consumers are the analysts, data scientists and engineers who look for products and request them. Providers are the organisations that list them. A provider can offer public data, free sample data or commercial data products, so a listing isn't necessarily free. Marketplace also matters for budgets: Databricks users can spend up to 10% of their Universal Commits on eligible third-party partner products bought through Marketplace.
Marketplace also includes Partner Connect integrations. Partner Connect lets you create trial accounts with selected Databricks technology partners and connect your workspace to their solutions from the Databricks UI. It sets up the Databricks resources a partner needs on your behalf and then passes those details to the partner. This is simpler than wiring up the connection by hand, and it lets you try a partner tool on your own lakehouse data before you adopt it.
Checkpoint 1 of 6· Check yourself
A team lead asks what Databricks Marketplace does. Which description is most accurate?
Marketplace is an exchange between outside providers and Databricks customers. Certifying tables and billing for compute belong to other parts of the platform.
“This gives customers a single catalog for finding data, software, and AI capabilities across the Databricks partner ecosystem.”Source: docs.databricks.com
Sources1
2.What a listing can contain
"Marketplace" sounds like it only sells datasets, but listings cover much more. The documentation lists datasets, AI models, notebooks, apps and Model Context Protocol (MCP) servers. Datasets usually arrive as catalogs of tabular data. Non-tabular data, such as images, audio or video, is also supported through Databricks volumes. When providers build a listing, they pick from these asset types:
| Asset type | What it is | Who can access it |
|---|---|---|
| Tables | Structured tabular data from a share | Any consumer, including third-party platforms |
| Files | Volumes holding data such as images, audio and video | Only consumers with a Unity Catalog-enabled Databricks workspace |
| Models | AI models that consumers can deploy for inference | Only consumers with a Unity Catalog-enabled Databricks workspace |
| Notebooks | Interactive notebooks for data science and ML workflows, often showing sample use cases | Only consumers with a Unity Catalog-enabled Databricks workspace |
| MCP server | Tools and APIs that agents use to interact with external systems | Only other users on Databricks can install them |
Two of these types need a closer look. Third-party apps are built by software vendors, consultants and data-sharing partners. Each app goes through a comprehensive security review before it is published. Once installed, an app runs inside your Databricks environment on your own compute, so your data stays in your workspace. MCP servers give agents tools for reaching external systems. In Databricks you can use them with Supervisor Agent and AI Playground. Databricks also recommends that providers include sample notebooks, which show example use cases. Consumers can preview a sample notebook and import it into their workspace.
Checkpoint 2 of 6· Match them up
Match each Marketplace listing type to what it gives a consumer
Tap a term, then the definition that fits it.
Each asset type does a different job. Volumes hold non-tabular data, apps run on your own compute, MCP servers serve agents, and notebooks demonstrate use cases.
“Third-party apps run within your Databricks environment and use your compute resources, so your data stays in your workspace.”Source: docs.databricks.com
Checkpoint 3 of 6· Exam question
A data team maintains a Delta table in Unity Catalog and wants partner organizations without Databricks accounts to query it directly, with updates appearing in near real time and no data copied into the recipient's environment. Which approach should the team use?
Correct answer: A — List the table on Databricks Marketplace, sharing it via OpenSharing for live, zero-copy access to the Delta table.
- A. Databricks Marketplace shares Delta and Iceberg tables through the OpenSharing protocol, which gives recipients live, zero-copy access without a Databricks account or a duplicated copy of the data.
- B. A nightly Parquet export creates a stale, duplicated copy of the data rather than the near-real-time, zero-copy access the requirement calls for.
- C. Requiring the partner to hold a full Databricks account and SQL warehouse defeats the goal of reaching organizations that have no Databricks account, and it bypasses Marketplace entirely.
- D. A daily emailed snapshot is a static, delayed artifact, not the live table access the scenario requires.
3.OpenSharing: the protocol underneath every listing
Marketplace doesn't copy files from provider to consumer. It uses OpenSharing for secure data exchange, and the docs say this protocol is what lets consumers access shared data securely, whether a listing is public or private.
The unit that OpenSharing shares is called a share. A share is a collection of tables, views, volumes and AI models that can be shared and secured as a single unit. A provider registers those assets in Unity Catalog, adds them to a share, and points a Marketplace listing at the share. This is why Unity Catalog appears in every requirement list: OpenSharing requires the provider's workspace to be enabled for Unity Catalog. It also explains the access column in the table above. Tables in a share can go to any consumer. Volumes, AI models and notebooks need a consumer workspace that is enabled for Unity Catalog.
Tables can be shared with any consumer, but volumes need a Unity Catalog-enabled Databricks workspace. A consumer without one gets only the tabular part of the share.
Checkpoint 4 of 6· Check yourself
A provider puts both tables and a volume in one share and lists it. What does a consumer without a Unity Catalog-enabled workspace receive?
Volumes are limited to consumers who have a Unity Catalog-enabled workspace. Everyone else can still read the tables in the share.
“include both tables and volumes in a share, consumers who don't have access to a Unity Catalog-enabled workspace can only access the tabular data.”Source: docs.databricks.com
4.Public listings and private exchanges
Every listing is either public or private. Every consumer can browse and view a public listing. A private listing lives in a private exchange, a group of invited member consumers set up by a provider. Only members can browse, view and request its listings. To add a member, the provider enters the member metastore's sharing identifier, a string in the format <cloud>:<region>:<uuid>. Members therefore have to be Databricks customers with a workspace attached to a Unity Catalog metastore.
Why use a private exchange instead of plain OpenSharing with the same partners? Members can discover data products before anything is shared with them, so they know what's available to request. They also get a storefront for requesting and accessing products, which may be easier to use than OpenSharing on its own. A private exchange can hold both instantly available listings and listings that need approval. A provider can also edit a public listing to make it private.
Checkpoint 5 of 6· Check yourself
What does a private exchange add on top of using OpenSharing directly with the same recipients?
The extra value of a private exchange is discovery and a storefront. Members must still have a Unity Catalog metastore, and payment never happens on Marketplace.
“private exchanges have the additional advantage of making data products discoverable by members before the data product is shared”Source: docs.databricks.com
Sources5
5.What it takes to publish a listing
To publish, a provider needs a Databricks account on the Premium plan or above and a workspace enabled for Unity Catalog. Not every workspace needs Unity Catalog; a provider can create one workspace just for managing listings. The provider must also agree to the Marketplace provider policies. Sign-up depends on where the listings will go. Providers who want public listings apply through the Databricks Data Partner Program. Providers who only want private exchanges can sign up themselves in the provider console, but that has to be done by a Databricks account admin. In both cases, a user needs the Marketplace admin role to open the Provider console.
When creating a listing, the provider chooses how consumers get access. Either the data assets are instantly accessible after the consumer accepts the terms, or consumers must request access and wait for approval. That choice can't be changed once the listing is saved as a draft or published. An instant listing must include a share when it's created. A by-request listing can leave the share out and add it later, after the business agreement is done and the request is approved.
Checkpoint 6 of 6· Check yourself
A provider is drafting a listing that requires approval because a contract has to be signed first. The share isn't ready yet. Can the listing be published?
Only free, instantly available listings need a share when they are created. By-request listings can add the share after the request is approved.
“Listings that require you to approve a consumer request, on the other hand, don't require that you include a share in the listing.”Source: docs.databricks.com
Exam traps
Each one states something that sounds right. Open it to see what is actually true.
1.Private exchange listings appear in the public Marketplace, and only access to them is restricted.Why is that wrong?
Private listings are hidden from the public marketplace entirely. Only members of the exchange can browse, view and request them.
Covered in Public listings and private exchanges
2.A provider can switch a listing between instant access and request-only at any time.Why is that wrong?
The access mode is fixed once the listing is saved as a draft or published.
Covered in What it takes to publish a listing
Sources
Every claim above is drawn from one of these pages, quoted as it was written on the date shown.
- 1.https://docs.databricks.com/aws/en/marketplaceOfficial docs
“This gives customers a single catalog for finding data, software, and AI capabilities across the Databricks partner ecosystem.”
↩︎ What Marketplace is for“You can share public data, free sample data, and commercialized data offerings.”
↩︎ What Marketplace is for“Databricks users can use up to 10% of their Universal Commits to purchase eligible third-party partner products through the Databricks Marketplace.”
↩︎ What Marketplace is for“Partner Connect lets you create trial accounts with select Databricks technology partners and connect your Databricks workspace to partner solutions from the Databricks UI.”
↩︎ What Marketplace is for“Listings include datasets, AI models, notebooks, apps, and Model Context Protocol (MCP) servers.”
↩︎ What a listing can contain“Datasets are typically made available as catalogs of tabular data, although non-tabular data, in the form of Databricks volumes, is also supported.”
↩︎ What a listing can contain“Marketplace uses OpenSharing for secure data exchange”
↩︎ OpenSharing: the protocol underneath every listing“Have a Databricks account and premium workspace that is enabled for Unity Catalog.”
↩︎ What it takes to publish a listing“Databricks Marketplace is an open exchange where data providers, software vendors, and technology partners publish offerings”
↩︎ Key concept - 2.
“the app goes through a comprehensive security review process before it can be published on Marketplace.”
↩︎ What a listing can contain“Third-party apps run within your Databricks environment and use your compute resources, so your data stays in your workspace.”
↩︎ Checkpoint - 3.
“A share is an OpenSharing object. It's a collection of tables, views, volumes, and AI models that are shareable and securable as a unit.”
↩︎ OpenSharing: the protocol underneath every listing“To list a data product that is free and instantly available to consumers, you must include a share when you create the listing.”
↩︎ What it takes to publish a listing“This field cannot be changed after the listing is saved as a draft or published.”
↩︎ Exam trap 2“include both tables and volumes in a share, consumers who don't have access to a Unity Catalog-enabled workspace can only access the tabular data.”
↩︎ Checkpoint“Listings that require you to approve a consumer request, on the other hand, don't require that you include a share in the listing.”
↩︎ Checkpoint - 4.
“OpenSharing, in turn, requires that your Databricks workspace is enabled for Unity Catalog.”
↩︎ OpenSharing: the protocol underneath every listing“To create public listings, you apply through the Databricks Data Partner Program.”
↩︎ What it takes to publish a listing“To access the Provider console, a user must have the Marketplace admin role.”
↩︎ What it takes to publish a listing - 5.
“Private exchange members must be Databricks customers and must have access to a workspace that is attached to a Unity Catalog metastore.”
↩︎ Public listings and private exchanges“provide consumers with a storefront interface for requesting and accessing data products that might be easier to use than OpenSharing on its own.”
↩︎ Public listings and private exchanges“Private exchange listings do not appear in the public marketplace.”
↩︎ Exam trap 1“Private exchange listings do not appear in the public marketplace.”
↩︎ Prediction“private exchanges have the additional advantage of making data products discoverable by members before the data product is shared”
↩︎ Checkpoint