Task 6.1: Identify trends in job performance using the Lakeflow Jobs run history view to compare current execution times against historical baselines.

• AThe Spark UI of the latest run's cluster

  Incorrect

  The Spark UI shows one cluster's jobs and stages; it has no history of earlier job runs.

• BThe compute event log of the job cluster

  Incorrect

  The event log records cluster events such as starts, resizes and terminations, not task durations across runs.

• CThe Timeline view of the latest job run

  Incorrect

  The Timeline view shows where time went inside one run; it does not compare against earlier runs.

• DThe matrix view on the job's Runs tab

  Correct

  The matrix view shows the history of runs with a cell per task, and bar heights indicate run durations, so trends across runs stand out.

The Runs tab's matrix view is the built-in history: each column is a run and each task's cell shows its status and duration.

• AThe matrix view on the job's Runs tab, scrolled back further

  Incorrect

  The Jobs UI keeps run history for 60 days, so runs from six months ago are no longer listed.

• BThe system.lakeflow.job_run_timeline system table

  Correct

  The jobs system tables record job runs across the account with a 365-day free retention period, and they can be queried for trends.

• CThe Spark UI of the job's most recent cluster

  Incorrect

  The Spark UI shows one cluster's Spark jobs and stages, not earlier job runs.

• DThe compute event log of the job cluster

  Incorrect

  The event log records cluster lifecycle events, not job run durations over months.

Use the Runs tab for recent history and the system.lakeflow tables for long-term analysis, such as duration trends against a baseline.

• ANothing, since Databricks keeps the details of each job run indefinitely

  Incorrect

  Databricks maintains job run history for up to 60 days.

• BSet a long timeout on the job so its runs stay open

  Incorrect

  A timeout limits how long a run may take; it does not extend how long results are kept.

• CClone the job each month so that each copy keeps its own run history

  Incorrect

  Run history is kept for up to 60 days whichever job a run belongs to, so the copies lose their older runs too.

• DExport each run's notebook results, such as to HTML, within 60 days

  Correct

  Databricks recommends exporting run results before they expire; notebook runs can be exported to HTML.

Run history in the Jobs UI lasts 60 days. Persist what you must keep: export notebook output or write results to tables.

• AThe run's Duration breakdown by phase

  Correct

  Hovering over a run's Duration shows the time in the Queued, Waiting for resources, Library installation and Running phases.

• BThe jobs timeline in the Spark UI of the job cluster

  Incorrect

  The Spark UI covers only what happens once Spark is running, not queueing or library installation.

• CThe compute event log of the cluster

  Incorrect

  The event log lists cluster events; it does not split a job run's duration into phases.

• DThe matrix view of the job's runs

  Incorrect

  The matrix view does not split a run into its queueing, waiting, installation and running phases.

The run breakdown shows which phase grew, so the fix can target queueing, compute start-up, library installs or the work itself.

• AThe matrix view on each job's page, opened one job at a time

  Incorrect

  The matrix covers a single job, so a workspace-wide picture would need every job opened in turn.

• BThe system.billing.usage table

  Incorrect

  The billing table records usage and cost, not run outcomes or error types.

• CThe Runs tab under Jobs & Pipelines, filtered to Jobs

  Correct

  It shows a Finished runs count graph and a Top 5 error types table for the selected time range across jobs.

• DFailure notifications configured on each job

  Incorrect

  Notifications report individual failures as they happen; they do not summarize error types.

The workspace-level Runs tab gives a cross-job view of recent outcomes and the most common errors, with filters for status, run as user and error code.