Skip to content

Data Engineer Professional · Domain 2: Data Ingestion & Acquisition · 12% of the exam

Task 2.5: Configure Lakehouse Federation with governance across supported source systems, applying UC permissions and connection-level credentials.

Connections and the credentials they hold, foreign catalogs, privileges on connections and foreign objects, pushdown, and catalog federation.

Study it

  • Lakehouse Federation: connections, foreign catalogs and permissions

    Lesson coming

Sample questions

Try each one before opening the answer. Every option is explained, with the Databricks documentation page that proves it.

Question 1 · choose 1

A platform team created a PostgreSQL connection and a foreign catalog pg_erp for it. Analysts must query only the sales schema of the database, through Unity Catalog, and must never see the database credentials or reach other schemas. Which grants follow least privilege?

  1. AUSE CONNECTION on the connection, plus USE CATALOG on pg_erp
  2. BMANAGE on the foreign catalog pg_erp for the analysts' group
  3. CUSE CATALOG on pg_erp; USE SCHEMA and SELECT on pg_erp.sales
  4. DUSE CATALOG, USE SCHEMA and SELECT on the whole pg_erp catalog
Show the answer and why
  • AUSE CONNECTION on the connection, plus USE CATALOG on pg_erp

    Incorrect

    USE CONNECTION shows the connection details and allows remote_query, which could reach schemas outside sales.

  • BMANAGE on the foreign catalog pg_erp for the analysts' group

    Incorrect

    MANAGE is an administrative privilege on the catalog, far beyond querying one schema.

  • CUSE CATALOG on pg_erp; USE SCHEMA and SELECT on pg_erp.sales

    Correct

    Foreign catalogs are governed like other catalogs; the connection's credentials are used on the analysts' behalf and are never exposed.

  • DUSE CATALOG, USE SCHEMA and SELECT on the whole pg_erp catalog

    Incorrect

    SELECT on the catalog is inherited by every schema in it, so analysts would reach all schemas.

A connection stores the path and credentials for the external database, and a foreign catalog mirrors a database for read-only queries governed by Unity Catalog privileges. USE CONNECTION is needed only to view the connection or to run remote_query.

Question 2 · choose 1

A data engineer must create a Unity Catalog connection to a MySQL database for Lakehouse Federation. Which privilege do they need if they are not a metastore admin?

  1. ACREATE CATALOG on the metastore
  2. BUSE CONNECTION on the metastore
  3. CCREATE CONNECTION on the metastore
  4. DCREATE EXTERNAL LOCATION on a storage credential
Show the answer and why
  • ACREATE CATALOG on the metastore

    Incorrect

    That is needed later for the foreign catalog, not for the connection.

  • BUSE CONNECTION on the metastore

    Incorrect

    USE CONNECTION applies to an existing connection.

  • CCREATE CONNECTION on the metastore

    Correct

    Creating a connection requires metastore admin or CREATE CONNECTION on the metastore.

  • DCREATE EXTERNAL LOCATION on a storage credential

    Incorrect

    That is for cloud storage, not database connections.

A connection specifies a path and credentials for an external database. Creating one needs metastore admin or CREATE CONNECTION on the metastore.

Question 3 · choose 1

A connection to PostgreSQL is owned by the platform team. A data engineer who already has CREATE CATALOG on the metastore must create a foreign catalog from it. What else do they need?

  1. AUSE CONNECTION on the PostgreSQL connection
  2. BMANAGE on the PostgreSQL database user
  3. CCREATE CATALOG on the connection
  4. DCREATE FOREIGN CATALOG on the connection
Show the answer and why
  • AUSE CONNECTION on the PostgreSQL connection

    Incorrect

    USE CONNECTION lets a user view the connection and run remote_query.

  • BMANAGE on the PostgreSQL database user

    Incorrect

    Database users are not Unity Catalog securables.

  • CCREATE CATALOG on the connection

    Incorrect

    CREATE CATALOG is a metastore privilege, which they already hold; it is not granted on connections.

  • DCREATE FOREIGN CATALOG on the connection

    Correct

    Creating a foreign catalog requires CREATE CATALOG on the metastore and either ownership of the connection or CREATE FOREIGN CATALOG on it.

To create a foreign catalog, a user needs CREATE CATALOG on the metastore and must own the connection or hold CREATE FOREIGN CATALOG on it.

Question 4 · choose 1

Analysts get an error when they query a foreign catalog from a classic SQL warehouse. Which warehouse types support Lakehouse Federation?

  1. AClassic SQL warehouses with Photon turned off
  2. BPro or serverless SQL warehouses
  3. CAny SQL warehouse with a larger cluster size
  4. DOnly serverless SQL warehouses
Show the answer and why
  • AClassic SQL warehouses with Photon turned off

    Incorrect

    Classic warehouses are not supported for federation.

  • BPro or serverless SQL warehouses

    Correct

    Query federation requires pro or serverless SQL warehouses at version 2023.40 or above.

  • CAny SQL warehouse with a larger cluster size

    Incorrect

    Size does not matter; the warehouse type does.

  • DOnly serverless SQL warehouses

    Incorrect

    Pro warehouses are also supported.

Lakehouse Federation needs Databricks Runtime 13.3 LTS or above with standard or dedicated access mode, or a pro or serverless SQL warehouse.

Question 5 · choose 1

A federated query against PostgreSQL filters rows, then ranks them with a window function. Which part does Lakehouse Federation not push down to PostgreSQL?

  1. AThe window function
  2. BThe WHERE filter
  3. CA LIMIT clause
  4. DComparison operators in the filter
Show the answer and why
  • AThe window function

    Correct

    The PostgreSQL pushdown table lists window functions as not supported.

  • BThe WHERE filter

    Incorrect

    Filters are pushed down on all compute.

  • CA LIMIT clause

    Incorrect

    Limits are pushed down on all compute.

  • DComparison operators in the filter

    Incorrect

    Comparison operators are pushed down.

Federation pushes supported operations, such as filters, limits and sorting with a limit, to PostgreSQL and runs the rest in Databricks. Window functions are not pushed down.

Practice domain 2 →Practice all domains →