Exam Bundle

That observation is a valid indicator for “Bronze ingest succeeds but silver quality checks suddenly drop many records” when using Inspect silver expectation metrics and rejected source samples. The scenario instead tests “A pipeline failure needs targeted replay without rerunning unaffected upstream tasks”; the relevant result is Only the intended failed task/path reruns and downstream outputs are correct.

A specific incompatible dependency/version is identified would demonstrate the separate action Review run environment versions and driver exception logs, which addresses “A production job fails only after a library was upgraded.” It does not verify the requirement “A pipeline failure needs targeted replay without rerunning unaffected upstream tasks” following Use supported Jobs task repair or selective pipeline retry where applicable.

Changed scan and shuffle costs explain the regression would demonstrate the separate action Inspect SQL query profile and pruning/scan metrics, which addresses “A lakehouse table read suddenly becomes slow after a data layout change.” It does not verify the requirement “A pipeline failure needs targeted replay without rerunning unaffected upstream tasks” following Use supported Jobs task repair or selective pipeline retry where applicable.

Review predictive optimization eligibility and query workload instead of assuming an error is appropriate when “A new liquid clustered table is created but no clustering keys were selected automatically”; it would be verified by The optimizer selects keys only when sufficient benefit is predicted. It does not directly satisfy “The team wants Databricks to manage table data lifecycle and optimization automatically,” whose relevant acceptance evidence is Maintenance history shows supported automatic optimization tasks for the managed table.

Use appropriately configured `VACUUM` with safe retention conditions belongs to the separate case “The table has huge amounts of obsolete files and operators need retention-aware cleanup,” where Unreferenced eligible files are removed without breaking required time travel would be expected. That differs from the condition in this stem: “The team wants Databricks to manage table data lifecycle and optimization automatically.”

The alternative Enable predictive optimization with supported workspace/metastore settings has a distinct purpose—Optimizations must run automatically for eligible managed Delta tables. It would not establish the required outcome (The team wants Databricks to manage table data lifecycle and optimization automatically); the necessary verification concerns Maintenance history shows supported automatic optimization tasks for the managed table.

This signal belongs to the adjacent case “A new liquid clustered table is created but no clustering keys were selected automatically,” whose remedy is Review predictive optimization eligibility and query workload instead of assuming an error. The observed result for the present requirement is Maintenance history shows supported automatic optimization tasks for the managed table.

That observation is a valid indicator for “Maintenance cost is rising on tables rarely queried” when using Use cost-aware optimization policies rather than blanket frequent rewrites. The scenario instead tests “The team wants Databricks to manage table data lifecycle and optimization automatically”; the relevant result is Maintenance history shows supported automatic optimization tasks for the managed table.

That observation is a valid indicator for “An external table is needed because a separate system must own and manage the storage path” when using Use a UC external table with a governed external location. The scenario instead tests “The team wants Databricks to manage table data lifecycle and optimization automatically”; the relevant result is Maintenance history shows supported automatic optimization tasks for the managed table.

Store the authoritative table in governed cloud object storage, not cache is appropriate when “An important dataset must remain queryable even after cached worker nodes terminate”; it would be verified by All table rows are readable from storage after a cluster restart. It does not directly satisfy “An external table is needed because a separate system must own and manage the storage path,” whose relevant acceptance evidence is The table references the existing owner-managed storage without moving it.

Enable appropriate predictive optimization features for supported managed tables belongs to the separate case “A managed table's storage optimization should reflect observed query patterns automatically,” where Optimization activity and improved data skipping follow the workload would be expected. That differs from the condition in this stem: “An external table is needed because a separate system must own and manage the storage path.”

Project only required columns and use a suitable columnar storage scan belongs to the separate case “A repeated SQL query's latency is dominated by materialization of unnecessary columns,” where Physical scan bytes decrease without changing output columns needed would be expected. That differs from the condition in this stem: “An external table is needed because a separate system must own and manage the storage path.”

That observation is a valid indicator for “A repeated SQL query's latency is dominated by materialization of unnecessary columns” when using Project only required columns and use a suitable columnar storage scan. The scenario instead tests “An external table is needed because a separate system must own and manage the storage path”; the relevant result is The table references the existing owner-managed storage without moving it.

Retries do not duplicate outputs and batch duration meets the SLA pertains to a different control (Make `foreachBatch` sink operations idempotent and right-size write throughput) and a different need (A streaming sink repeatedly retries due to expensive non-idempotent external writes). It cannot establish that Use a UC external table with a governed external location met the stated goal.

Optimization activity and improved data skipping follow the workload would demonstrate the separate action Enable appropriate predictive optimization features for supported managed tables, which addresses “A managed table's storage optimization should reflect observed query patterns automatically.” It does not verify the requirement “An external table is needed because a separate system must own and manage the storage path” following Use a UC external table with a governed external location.

This action solves a different issue: “An external table is needed because a separate system must own and manage the storage path.” Its normal signal is The table references the existing owner-managed storage without moving it. Here the requested behavior is “A managed table's storage optimization should reflect observed query patterns automatically,” so that signal is not sufficient.

Review predictive optimization eligibility and query workload instead of assuming an error belongs to the separate case “A new liquid clustered table is created but no clustering keys were selected automatically,” where The optimizer selects keys only when sufficient benefit is predicted would be expected. That differs from the condition in this stem: “A managed table's storage optimization should reflect observed query patterns automatically.”

Use `CLUSTER BY AUTO` with required predictive optimization support belongs to the separate case “A UC managed table should choose clustering columns from observed workloads,” where Selected keys adapt to query patterns over time where beneficial would be expected. That differs from the condition in this stem: “A managed table's storage optimization should reflect observed query patterns automatically.”

The table references the existing owner-managed storage without moving it pertains to a different control (Use a UC external table with a governed external location) and a different need (An external table is needed because a separate system must own and manage the storage path). It cannot establish that Enable appropriate predictive optimization features for supported managed tables met the stated goal.

Selected keys adapt to query patterns over time where beneficial would demonstrate the separate action Use `CLUSTER BY AUTO` with required predictive optimization support, which addresses “A UC managed table should choose clustering columns from observed workloads.” It does not verify the requirement “A managed table's storage optimization should reflect observed query patterns automatically” following Enable appropriate predictive optimization features for supported managed tables.

That observation is a valid indicator for “Optimizations must run automatically for eligible managed Delta tables” when using Enable predictive optimization with supported workspace/metastore settings. The scenario instead tests “A managed table's storage optimization should reflect observed query patterns automatically”; the relevant result is Optimization activity and improved data skipping follow the workload.

The alternative Run supported rewrite/maintenance plus appropriate retention-safe VACUUM where required has a distinct purpose—Physical files contain logically deleted rows and must eventually be purged under compliance policy. It would not establish the required outcome (The table has huge amounts of obsolete files and operators need retention-aware cleanup); the necessary verification concerns Unreferenced eligible files are removed without breaking required time travel.

This action solves a different issue: “The table's retention setting conflicts with consumers needing older time travel snapshots.” Its normal signal is Required historical versions remain queryable throughout retention window. Here the requested behavior is “The table has huge amounts of obsolete files and operators need retention-aware cleanup,” so that signal is not sufficient.

Create it in the intended UC catalog and schema as a managed table is appropriate when “A new curated dataset requires centralized ownership and consistent permission inheritance”; it would be verified by The intended owner and inherited privileges apply to the new dataset. It does not directly satisfy “The table has huge amounts of obsolete files and operators need retention-aware cleanup,” whose relevant acceptance evidence is Unreferenced eligible files are removed without breaking required time travel.

This signal belongs to the adjacent case “A new curated dataset requires centralized ownership and consistent permission inheritance,” whose remedy is Create it in the intended UC catalog and schema as a managed table. The observed result for the present requirement is Unreferenced eligible files are removed without breaking required time travel.

Required historical versions remain queryable throughout retention window pertains to a different control (Set retention policy based on documented recovery requirements before cleanup) and a different need (The table's retention setting conflicts with consumers needing older time travel snapshots). It cannot establish that Use appropriately configured `VACUUM` with safe retention conditions met the stated goal.

That observation is a valid indicator for “A relatively stable very low-cardinality partition key is required by legacy consumers” when using Use a supported partition strategy with a small manageable partition count. The scenario instead tests “The table has huge amounts of obsolete files and operators need retention-aware cleanup”; the relevant result is Unreferenced eligible files are removed without breaking required time travel.

This action solves a different issue: “An important dataset must remain queryable even after cached worker nodes terminate.” Its normal signal is All table rows are readable from storage after a cluster restart. Here the requested behavior is “A new curated dataset requires centralized ownership and consistent permission inheritance,” so that signal is not sufficient.

This action solves a different issue: “A UC managed table should choose clustering columns from observed workloads.” Its normal signal is Selected keys adapt to query patterns over time where beneficial. Here the requested behavior is “A new curated dataset requires centralized ownership and consistent permission inheritance,” so that signal is not sufficient.

Use a UC external table with a governed external location is appropriate when “An external table is needed because a separate system must own and manage the storage path”; it would be verified by The table references the existing owner-managed storage without moving it. It does not directly satisfy “A new curated dataset requires centralized ownership and consistent permission inheritance,” whose relevant acceptance evidence is The intended owner and inherited privileges apply to the new dataset.

That observation is a valid indicator for “A batch pipeline writes too many output files for its final dataset size” when using Adjust output partitioning before writing within safe memory limits. The scenario instead tests “A new curated dataset requires centralized ownership and consistent permission inheritance”; the relevant result is The intended owner and inherited privileges apply to the new dataset.

That observation is a valid indicator for “The table has huge amounts of obsolete files and operators need retention-aware cleanup” when using Use appropriately configured `VACUUM` with safe retention conditions. The scenario instead tests “A new curated dataset requires centralized ownership and consistent permission inheritance”; the relevant result is The intended owner and inherited privileges apply to the new dataset.

This signal belongs to the adjacent case “A new liquid clustered table is created but no clustering keys were selected automatically,” whose remedy is Review predictive optimization eligibility and query workload instead of assuming an error. The observed result for the present requirement is The intended owner and inherited privileges apply to the new dataset.