Skip to content
Cheat sheets

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

Task 2.4: Implement OpenSharing (D2D and Databricks-to-Open) and Clean Rooms for privacy-preserving collaboration.

Shares, recipients and providers, Databricks-to-Databricks and open sharing, history and partitions in shares, and clean rooms with notebooks and output tables.

Study it

  • OpenSharing and Clean Rooms

    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 provider shares the table sales.orders with a partner's Databricks workspace through Databricks-to-Databricks OpenSharing. The share was created years ago without table history. The partner now wants to process only new and changed rows from the change data feed instead of reading the full table each day. What must the provider do?

  1. AEnable change data feed, then share the table WITH HISTORY
  2. BGrant the recipient MODIFY on the share as well as SELECT
  3. CCreate a new open sharing recipient with a bearer token for the partner
  4. DAdd the table to the share again under an alias with a partition filter
Show the answer and why

  • AEnable change data feed, then share the table WITH HISTORY

    Correct

    Recipients can read the change data feed only if CDF was enabled on the table before it was shared with history.

  • BGrant the recipient MODIFY on the share as well as SELECT

    Incorrect

    Recipients get read access through SELECT on a share; MODIFY is not a share privilege and does not expose changes.

  • CCreate a new open sharing recipient with a bearer token for the partner

    Incorrect

    The protocol is not the problem; without history and CDF no recipient can query the changes.

  • DAdd the table to the share again under an alias with a partition filter

    Incorrect

    Re-adding the table under an alias or with a filter does not make its changes readable; that depends on history and the change data feed.

A recipient can read the change data feed of a shared table only when the provider enabled change data feed on the table before sharing it WITH HISTORY. A share made without history, or a table whose change data feed was off when it was shared, gives the recipient no change records.

Databricks documentation

Task 2.4Report a problem with this question

Question 2 · choose 1

In a two-party clean room, a data engineer uploads a notebook that joins both parties' tables. What must happen before the notebook can run?

  1. AThe creator of the clean room must grant it EXECUTE
  2. BA Databricks admin must sign off on the notebook
  3. CThe other collaborator must approve the notebook
  4. DNothing; the uploader can run it at once
Show the answer and why

  • AThe creator of the clean room must grant it EXECUTE

    Incorrect

    All collaborators have equal privileges; approval is required.

  • BA Databricks admin must sign off on the notebook

    Incorrect

    Approval comes from the collaborators, not Databricks.

  • CThe other collaborator must approve the notebook

    Correct

    Clean rooms require approval by the other collaborators before a notebook can run.

  • DNothing; the uploader can run it at once

    Incorrect

    A notebook must be approved by others before it runs.

Clean rooms follow a no-trust model in which every collaborator, including the creator, has equal privileges. A notebook must be approved by the other collaborators before it can run.

Databricks documentation

Task 2.4Report a problem with this question

Question 3 · choose 1

Three companies run a clean room. A fourth company wants to join next month. What does the clean room design allow?

  1. AIt cannot join; clean rooms are locked to new collaborators
  2. BIt can join if all three companies vote to add it to the room
  3. CIt can join as long as the clean room stays under 100 members
  4. DIt can join once the creator renames and republishes the clean room
Show the answer and why

  • AIt cannot join; clean rooms are locked to new collaborators

    Correct

    After creation, a clean room is locked down so new collaborators cannot join.

  • BIt can join if all three companies vote to add it to the room

    Incorrect

    There is no process to add collaborators to an existing clean room.

  • CIt can join as long as the clean room stays under 100 members

    Incorrect

    Each clean room is limited to ten collaborators, and none can be added later.

  • DIt can join once the creator renames and republishes the clean room

    Incorrect

    Clean rooms cannot be renamed.

Each clean room is limited to ten collaborators, it cannot be renamed, and after creation it is locked so that no new collaborators can join.

Databricks documentation

Task 2.4Report a problem with this question

Question 4 · choose 1

A provider shortens the metastore's default recipient token lifetime from one year to 90 days. Existing Databricks-to-Open recipients still have tokens that expire next year. What must the provider do?

  1. ANothing; the new lifetime applies to all tokens at once
  2. BRotate the existing recipients' tokens
  3. CDrop and recreate the share
  4. DSwitch the recipients to Databricks-to-Databricks sharing
Show the answer and why

  • ANothing; the new lifetime applies to all tokens at once

    Incorrect

    Existing recipient tokens keep their lifetime.

  • BRotate the existing recipients' tokens

    Correct

    The new default does not update existing recipients; their tokens must be rotated to get the new lifetime.

  • CDrop and recreate the share

    Incorrect

    Tokens belong to recipients, not shares.

  • DSwitch the recipients to Databricks-to-Databricks sharing

    Incorrect

    Those recipients have no Databricks workspace to use it.

Recipient tokens are used only for Databricks-to-Open sharing and last at most one year. Changing the default lifetime does not change existing recipients; rotate their tokens to apply it.

Databricks documentation

Task 2.4Report a problem with this question

Question 5 · choose 1

A provider shares a Delta table that has deletion vectors enabled with a Databricks-to-Databricks recipient, with history. The recipient's queries from a cluster on Databricks Runtime 13.3 LTS fail. How can the recipient query the table?

  1. AAsk the provider to add the table to the share again under an alias
  2. BAsk the provider to add the table with a partition specification
  3. CAsk the provider to share the table WITHOUT HISTORY instead
  4. DQuery it from a SQL warehouse or Databricks Runtime 14.1 or above
Show the answer and why

  • AAsk the provider to add the table to the share again under an alias

    Incorrect

    An alias changes the name that recipients see, not how they read it.

  • BAsk the provider to add the table with a partition specification

    Incorrect

    A partition specification limits the shared rows, not the reader needed.

  • CAsk the provider to share the table WITHOUT HISTORY instead

    Incorrect

    A table with deletion vectors or column mapping must be shared with history.

  • DQuery it from a SQL warehouse or Databricks Runtime 14.1 or above

    Correct

    Recipients query tables with deletion vectors or column mapping from a SQL warehouse, Databricks Runtime 14.1 or above, or delta-sharing-spark 3.1 or above.

Tables with deletion vectors or column mapping are shared with history. Recipients can query them from a SQL warehouse, from compute on Databricks Runtime 14.1 or above, or from compute running open source delta-sharing-spark 3.1 or above.

Databricks documentation

Task 2.4Report a problem with this question

Practice domain 2 →Practice all domains →