Overview
This section is worth 11% of the exam and covers three objectives. The first asks you to describe the platform's core components: Mosaic AI, Lakeflow Declarative Pipelines (formerly Delta Live Tables), Lakeflow Jobs, the Data Intelligence Engine, Delta Lake, Lakebase, Unity Catalog and Databricks SQL. The second moves into Catalog Explorer, where you browse catalogs and schemas, tell managed tables from external ones, read access controls, and recognise views, certified tables, metric views and lineage. The third covers Databricks Marketplace as a place to find and install shared data products.
Databricks refreshed the exam guide in October 2026 (the file is labelled August 2026) and the new version went live on 19 October 2026. Lakebase, metric views and the renamed Genie Agents are new to the outline, so older study material is out of date. The exam still has 45 scored questions in 90 minutes, and Databricks does not publish a passing score.
How it's tested
Expect short descriptive stems. A capability or business need is described and you name the component, the Unity Catalog object or the Marketplace feature that provides it. Several options will be real Databricks products, so the skill is matching purpose precisely rather than spotting a fake name.
-
Name the component
A job such as orchestrating tasks, serving low-latency transactional data or building ML models is described, and you choose the platform component responsible.
-
Read Catalog Explorer
A screenshot-style description of an object's tabs or badges asks whether it is a managed table, external table, view, metric view or certified asset.
-
Marketplace fit
A team wants third-party or partner data without building pipelines, and you decide whether Marketplace and Delta Sharing meet the need.
A finance team wants one governed definition of net revenue that dashboards and Genie both reuse, so totals stop differing between reports. A regular view could hold the SQL, but every consumer would still aggregate it differently. A metric view stores the measure and its dimensions in Unity Catalog and is queried with the MEASURE function, which keeps calculations consistent. The metric view is the stronger answer.
Objectives 3
Official objectives from the Databricks outline, in outline order.
1.1 Describe the core components of the Databricks Data Intelligence Platform (Lakehouse Platform), including Mosaic AI, Lakeflow Declarative Pipelines (DeltaLive tables), Lakeflow Jobs, Data Intelligence Engine, Delta Lake, Lakebase, Unity Catalog, and Databricks SQL.The purpose of each named platform component.
- What it means
- The Data Intelligence Platform is the lakehouse plus AI features. Delta Lake is the open table format underneath, Unity Catalog governs every asset, Databricks SQL is the warehousing and BI layer, Lakeflow Declarative Pipelines builds managed ETL, Lakeflow Jobs orchestrates tasks, Mosaic AI covers model building and serving, Lakebase is a managed Postgres database for operational workloads, and the Data Intelligence Engine applies generative AI to understand your data's semantics.
- What to understand
- Learn one sentence of purpose for each component and the boundary between neighbours. Pipelines declare transformations and Databricks manages their dependencies, while Jobs schedules and sequences any kind of task, including pipelines, notebooks and SQL. Delta Lake handles analytical tables with ACID transactions and time travel; Lakebase handles row-level transactional reads and writes for applications. The Data Intelligence Engine is what powers Genie, Genie Code and AI-generated comments.
- How it shows up
- A one-line requirement is given and four genuine component names are offered; only one matches the stated purpose.
- Common confusion
- Picking Lakeflow Declarative Pipelines for a task that only needs scheduling, or treating Lakebase as a Delta table store.
1.2 Understand catalogs, schemas, managed and external tables, access controls, views, certified tables, metric views, and lineage within the Catalog Explorer interface.Unity Catalog objects as they appear in Catalog Explorer.
- What it means
- Catalog Explorer is the UI for browsing Unity Catalog. It shows catalogs and their schemas, the tables, views, volumes, functions and models inside them, and tabs for each object's columns, sample data, details, permissions, history and lineage. Badges or tags mark certified (endorsed) assets, and metric views appear as their own object type.
- What to understand
- A managed table's files live in storage that Unity Catalog controls, so dropping it removes the data; an external table points to a path you manage, so dropping it only removes metadata. A view stores a query, not data. A metric view defines measures and dimensions in YAML over a source table so that business metrics are calculated the same way everywhere. Lineage shows upstream and downstream tables, columns, notebooks, jobs and dashboards, captured automatically from queries run on Unity Catalog.
- How it shows up
- Questions describe what you see on an object's Details, Permissions or Lineage tab and ask what kind of object it is or what it implies.
- Common confusion
- Assuming lineage must be documented manually, when Unity Catalog records it from executed queries.
1.3 Describe the role and features of Databricks Marketplace.Finding and installing shared data products.
- What it means
- Databricks Marketplace is an open exchange for data products such as datasets, notebooks, solution accelerators and ML models. Providers list products, and consumers browse, request or install them. Delivery runs on Delta Sharing, so shared data is queried in place rather than copied through an export process.
- What to understand
- Installed listings appear as a read-only catalog in Unity Catalog, which means they are governed and permissioned like any other catalog. Listings can be free and instantly available or require a provider to approve the request. Private exchanges let a provider offer listings to selected consumers only. Marketplace is open, so consumers do not strictly need Databricks if the listing uses open sharing.
- How it shows up
- A team wants external reference data quickly and with no pipeline; Marketplace with Delta Sharing is usually the expected answer.
- Common confusion
- Thinking Marketplace data is copied into your storage, or that installed shares can be modified.
High-yield concepts
The terms, rules and formulas this domain turns into questions most often.
-
Data Intelligence Platform
Lakehouse architecture enhanced with generative AI features.
- Remember
- Unity Catalog and Delta Lake sit underneath everything.
- Tested as
- Framing for component questions.
- Trap
- Treating it as only a Spark cluster service.
-
Delta Lake
Open table format adding ACID transactions to Parquet files.
- Remember
- Default format for Databricks tables.
- Tested as
- Time travel and history questions.
- Trap
- Confusing it with the catalog layer.
-
Unity Catalog
Unified governance for data and AI assets.
- Remember
- Holds permissions, lineage, tags and discovery.
- Tested as
- Appears in most sections.
- Trap
- Thinking it stores the table files itself.
-
Databricks SQL
Warehousing and BI layer: SQL editor, warehouses, dashboards, alerts.
- Remember
- Runs on SQL warehouses.
- Tested as
- Core analyst tooling.
- Trap
- Confusing a SQL warehouse with a cluster for notebooks.
-
Lakeflow Declarative Pipelines
Declarative ETL formerly called Delta Live Tables.
- Remember
- Builds streaming tables and materialized views.
- Tested as
- Component identification.
- Trap
- Using it as a scheduler.
-
Lakeflow Jobs
Orchestration of notebooks, SQL, pipelines and other tasks.
- Remember
- Triggers, schedules, retries and dependencies.
- Tested as
- Component identification.
- Trap
- Mixing it up with pipelines.
-
Lakebase
Fully managed Postgres-compatible operational database.
- Remember
- For low-latency transactional workloads.
- Tested as
- New to the 2026 outline.
- Trap
- Calling it a Delta Lake feature.
-
Mosaic AI
Tools to build, evaluate and serve ML and generative AI.
- Remember
- Model serving and agent building live here.
- Tested as
- Component identification.
- Trap
- Choosing it for BI reporting.
-
Metric view
Unity Catalog object defining reusable measures and dimensions.
- Remember
- Queried with the MEASURE function.
- Tested as
- Semantic consistency questions.
- Trap
- Treating it as an ordinary view.
-
Certified table
Asset marked as trusted by its owners.
- Remember
- Signals the approved source for analysis.
- Tested as
- Discovery questions.
- Trap
- Assuming certification grants access.
-
Databricks Marketplace
Exchange for datasets, notebooks, models and accelerators.
- Remember
- Delivered through Delta Sharing.
- Tested as
- Data acquisition scenarios.
- Trap
- Expecting to edit installed data.
Side by side
Concepts candidates mix up, compared on the point that decides the answer.
Platform components often confused
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| Lakeflow Declarative Pipelines | Declarative ETL with managed dependencies and data quality expectations. | Defines how data is transformed. | Picked for plain scheduling needs. |
| Lakeflow Jobs | Orchestrator that runs and schedules tasks. | Defines when and in what order work runs. | Assumed to transform data on its own. |
| Delta Lake | Analytical table format on cloud object storage. | Optimised for scans and large aggregates. | Chosen for high-frequency row updates from an app. |
| Lakebase | Managed Postgres database inside the platform. | Optimised for transactional reads and writes. | Mistaken for a Delta Lake feature. |
Objects in Catalog Explorer
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| Managed table | Data stored in Unity Catalog managed storage. | DROP removes metadata and data. | Thinking files survive a drop. |
| External table | Data stored at a path you register. | DROP removes only the metadata. | Assuming Databricks deletes the files. |
| View | Saved query with no stored data. | Recomputes at read time. | Expecting faster reads than the base table. |
| Metric view | YAML-defined measures and dimensions. | Aggregation logic travels with the object. | Querying it with plain SUM instead of MEASURE. |
Exam traps
Each trap is an answer that looks right. Learn the clue that should change your mind.
-
Pipelines versus Jobs
- What candidates pick
- Lakeflow Declarative Pipelines for a nightly notebook run.
- Why it looks right
- Both carry the Lakeflow name.
- Why it's wrong
- A nightly notebook needs scheduling and retries, which is the orchestrator's role; pipelines define transformations rather than scheduling arbitrary tasks.
Clue to watch for Words like schedule, trigger, retry or task order.
-
Lakebase as a table format
- What candidates pick
- Delta Lake when an app needs millisecond row lookups.
- Why it looks right
- Delta Lake is the best-known storage answer.
- Why it's wrong
- Delta tables suit analytical scans, while Lakebase provides a Postgres engine designed for operational transactions.
Clue to watch for Application, transactional, low latency or Postgres.
-
Dropping an external table
- What candidates pick
- The underlying files are deleted.
- Why it looks right
- Managed tables behave that way.
- Why it's wrong
- For external tables Unity Catalog only removes the registration; the files at the external location stay put.
Clue to watch for The table was created with a LOCATION path.
-
Metric view versus view
- What candidates pick
- Create a standard view with the revenue formula.
- Why it looks right
- Views are familiar and easy.
- Why it's wrong
- A plain view leaves aggregation to each consumer, so totals can still diverge; metric views carry the measure definition itself.
Clue to watch for Consistent metric definitions across tools.
-
Marketplace copying
- What candidates pick
- Marketplace exports a copy into your bucket.
- Why it looks right
- Many data vendors deliver files.
- Why it's wrong
- Marketplace uses Delta Sharing, so consumers query the provider's live data without copying it.
Clue to watch for No ingestion pipeline, always current.
-
Certified means accessible
- What candidates pick
- Certification lets everyone read the table.
- Why it looks right
- Certified sounds like approved for all.
- Why it's wrong
- Certification is a trust signal only; reading still requires USE CATALOG, USE SCHEMA and SELECT grants.
Clue to watch for A user sees the badge but gets permission denied.
Answer framework
Use these steps when a stem asks which platform component or catalog object fits a requirement.
- Classify the need as storage, governance, transformation, orchestration, BI, AI or operational database before reading the options. Many wrong options are real features that solve a different class of problem.
- Map the class to one component: Delta Lake, Unity Catalog, Declarative Pipelines, Jobs, Databricks SQL, Mosaic AI or Lakebase.
- For catalog questions, ask whether data is stored, where it lives, and whether a drop would delete files. Dropping a managed table eventually removes its data files, while dropping an external table leaves the files in place.
- If the stem mentions consistent business metrics, prefer a metric view; if it mentions trusted sources, think certification.
- For outside data with no pipeline, check whether Marketplace and Delta Sharing satisfy the requirement before choosing ingestion.
Practice question
Want more like this? The Databricks DA Assoc practice test has questions for every domain.
Checklist
0 of 11
Tick each one you can do without notes. Your ticks stay in this browser.
Study strategy
Start with a component map. On one page, write each of the eight named components with a single sentence of purpose and one example workload. Add a column for the component it is most often confused with. This page alone covers most of the questions this section produces.
Next, spend an hour in Catalog Explorer in a workspace you can use, ideally the Databricks Free Edition. Open the samples catalog, look at a table's Details tab to see whether it is managed or external, check the Permissions tab, and open the Lineage tab after running a query that creates a new table from it. Create a small managed table and an external one if you have an external location, drop both, and observe the difference. If metric views are available to you, create one from a sales table and query it with MEASURE.
Finally, browse Marketplace, open a free listing and install it. Note that it appears as a catalog you cannot write to. Because the October 2026 guide added Lakebase and metric views, read the current product documentation pages for those two rather than relying on older courses. Review your component map again the day before the exam.
FAQ
How much of the Databricks Data Analyst exam is Understanding of Databricks Data Intelligence Platform?
This section carries 11% of the exam, which is roughly five of the 45 scored questions. The weight comes from the official certification page and the current exam guide.
Which exam guide version is current?
The guide labelled August 2026 was updated on 8 October 2026, and that exam version went live on 19 October 2026. Databricks asks candidates to recheck the guide two weeks before their test date.
Do I need to know Mosaic AI in depth?
No. Analysts need to know what Mosaic AI is for, building and serving ML and generative AI models, so they can rule it in or out of a scenario.
What is Lakebase?
Lakebase is a managed Postgres-compatible database for operational applications, integrated with the rest of the platform. It appears in the October 2026 outline as a named core component.
Is Delta Live Tables still on the exam?
The capability is, under its current name Lakeflow Declarative Pipelines. The guide shows the old name in brackets, so recognise both.
What is the passing score?
Databricks does not publish a passing score for this exam. Results are reported as pass or fail.
Are metric views tested?
Yes. The Catalog Explorer objective now lists metric views alongside tables, views, certified tables and lineage.
Other Data & Analytics certifications
Where the same subject shows up in other exams. Cards marked "Same subject" open the matching domain.
Test your Domain 1 knowledge
Check that you can match every platform component, Catalog Explorer object and Marketplace feature to its purpose before moving on to the larger query sections.
Keep going
- Training Training Built on your Weak areas
- Practice test Questions by domain, with explanations
- Mock exam Full length, timed like the real exam
- Readiness test Find your weakest domain in minutes
- Cheat sheet Key terms and formulas on one page
- Ask the Databricks DA Assoc AI tutor Get any concept explained your way