Skip to content

Data Engineer Professional · Domain 4: Monitoring and Alerting · 10% of the exam

Task 4.4: Use Databricks Lakehouse alerts to monitor governed business metrics, data quality and anomaly signals, usage and cost, SQL warehouse/query health, audit and security events, AI agent quality, and Lakeflow Job branching.

Databricks SQL alerts, their conditions and schedules, notifications, alert tasks in jobs, and alerts over system tables and data quality results.

Study it

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 Databricks SQL alert checks for negative inventory every hour on its own schedule. The data team also needs it evaluated right after the nightly load job finishes, which happens at a different time each night, with the result sent to the load team. The hourly schedule must keep running. What should they do?

  1. AChange the alert's schedule to a cron time after the usual end of the load, notifying the load team
  2. BAdd a SQL task that runs the alert's query after the load task
  3. CAdd a SQL alert task that points to a legacy alert copied from it
  4. DAdd a SQL alert task after the load task, with the load team as subscribers
Show the answer and why
  • AChange the alert's schedule to a cron time after the usual end of the load, notifying the load team

    Incorrect

    The load ends at a different time each night, and this replaces the hourly schedule.

  • BAdd a SQL task that runs the alert's query after the load task

    Incorrect

    A SQL task runs the query but does not evaluate the alert condition or send the alert's notifications.

  • CAdd a SQL alert task that points to a legacy alert copied from it

    Incorrect

    SQL alert tasks support only Databricks SQL alerts, not legacy alerts.

  • DAdd a SQL alert task after the load task, with the load team as subscribers

    Correct

    The task evaluates the alert independently of its schedule and notifies the subscribers configured on the task.

A SQL alert task runs the alert's query on a pro or serverless warehouse, evaluates its condition and notifies the task's subscribers. It evaluates the alert independently of any schedule on the alert, which is not affected by the job run.

Question 2 · choose 1

A job's SQL alert task evaluates an alert whose condition is met. What status does the task report?

  1. AFailed, because the alert condition was met
  2. BSucceeded, because the alert was evaluated
  3. CSkipped, because notifications were sent
  4. DSucceeded with failures
Show the answer and why
  • AFailed, because the alert condition was met

    Incorrect

    Failed is reported for evaluation errors such as query failures.

  • BSucceeded, because the alert was evaluated

    Correct

    The task reports Succeeded when evaluation works, whether or not the condition triggered.

  • CSkipped, because notifications were sent

    Incorrect

    Notification delivery does not skip the task.

  • DSucceeded with failures

    Incorrect

    That status describes job runs where some tasks failed.

A SQL alert task reports Succeeded when the alert was evaluated, regardless of the condition, and Failed when evaluation hit an error, such as a warehouse connectivity issue or a query error.

Question 3 · choose 1

A team wants a job task that checks a threshold with a query that takes the job's region parameter. Why does a SQL alert task not fit?

  1. ASQL alert tasks run only on Classic SQL warehouses
  2. BSQL alert tasks cannot send notifications
  3. CSQL alert tasks can evaluate only alerts that the job owner created
  4. DSQL alert tasks do not support parameters, so a SQL task is needed
Show the answer and why
  • ASQL alert tasks run only on Classic SQL warehouses

    Incorrect

    They need a serverless or pro warehouse.

  • BSQL alert tasks cannot send notifications

    Incorrect

    They notify their subscribers.

  • CSQL alert tasks can evaluate only alerts that the job owner created

    Incorrect

    The requirement is CAN RUN permission on the alert, not ownership.

  • DSQL alert tasks do not support parameters, so a SQL task is needed

    Correct

    For parameterized queries, Databricks suggests a SQL task instead.

SQL alert tasks do not support parameters and accept only Databricks SQL alerts, not legacy alerts. For parameterized queries, use a SQL task.

Question 4 · choose 1

A team migrating from legacy alerts wants to point a new Databricks SQL alert at an existing saved query. What does the current alert design allow?

  1. AAny saved query can be attached, as with legacy alerts
  2. BOnly queries over metric views can be attached
  3. CEach alert owns its query, written in the alert editor
  4. DThe alert can use the query only if its state is UNKNOWN
Show the answer and why
  • AAny saved query can be attached, as with legacy alerts

    Incorrect

    Query reuse is one of the documented differences from legacy alerts.

  • BOnly queries over metric views can be attached

    Incorrect

    Alerts can query metric views, but they still own their query.

  • CEach alert owns its query, written in the alert editor

    Correct

    An existing saved query cannot be reused; each alert owns its query definition.

  • DThe alert can use the query only if its state is UNKNOWN

    Incorrect

    The UNKNOWN state no longer exists; alerts resolve to OK, TRIGGERED or ERROR.

The latest Databricks SQL alerts own their query definitions and resolve to OK, TRIGGERED or ERROR, without the UNKNOWN status of legacy alerts.

Question 5 · choose 1

Finance wants an email when yesterday's serverless DBUs exceed a set amount. Which approach fits?

  1. AA Databricks SQL alert on a query of system.billing.usage
  2. BA pipeline expectation on a copy of system.billing.usage
  3. CA CHECK constraint on system.billing.usage
  4. DA job duration threshold notification on the serverless jobs
Show the answer and why
  • AA Databricks SQL alert on a query of system.billing.usage

    Correct

    Alerts can monitor usage and cost by querying system tables on a schedule.

  • BA pipeline expectation on a copy of system.billing.usage

    Incorrect

    Expectations check data in pipeline datasets; they do not send cost alerts.

  • CA CHECK constraint on system.billing.usage

    Incorrect

    System tables are read-only; constraints cannot be added to them.

  • DA job duration threshold notification on the serverless jobs

    Incorrect

    Duration thresholds watch run time, not spend.

Databricks SQL alerts monitor any query result on a schedule, including queries over system tables for serverless billing or ingestion cost.

Practice domain 4 →Practice all domains →