Implementing Data Engineering Solutions Using Azure Databricks DP-750 Practice Test 14
Question 130
For this Microsoft DP-750 case: A business glossary has table names but no indication of which field is net revenue versus gross sales. Which design or operational response BEST fits?
- Implement a catalog stewardship process to review metadata and governed tags at dataset publication.
- Add reviewed Unity Catalog table and column comments, owners and business definitions for consumer discovery.
- Reconcile catalog object identity, lineage and current grants after schema or owner changes.
- Develop an approved role-aware column mask and test its outputs across supported query clients.
Explanation
Correct Answer
B:The most appropriate action is: Add reviewed Unity Catalog table and column comments, owners and business definitions for consumer discovery. It addresses the stated condition: A business glossary has table names but no indication of which field is net revenue versus gross sales. The technical basis is: Governed descriptive metadata reduces ambiguity and supports correct analytical interpretation. Without this control, analysts may combine incompatible metrics and deliver unreliable reports. A direct acceptance check is: A data steward can find the certified revenue definition and verify the correct field in Catalog Explorer.
Incorrect Answers
A:This alternative primarily addresses a different Azure Databricks problem: A central repository contains schemas with missing column owners and no sensitivity classifications. The corresponding measure there would be: Implement a catalog stewardship process to review metadata and governed tags at dataset publication. The stated situation instead calls for: Add reviewed Unity Catalog table and column comments, owners and business definitions for consumer discovery.
C:This alternative primarily addresses a different Azure Databricks problem: A governance team grants dataset access solely by a table's displayed name while records are moved to a new schema. The corresponding measure there would be: Reconcile catalog object identity, lineage and current grants after schema or owner changes. The stated situation instead calls for: Add reviewed Unity Catalog table and column comments, owners and business definitions for consumer discovery.
D:This alternative primarily addresses a different Azure Databricks problem: A data product stores PII in a shared Delta table but several reports require different masking for each audience. The corresponding measure there would be: Develop an approved role-aware column mask and test its outputs across supported query clients. The stated situation instead calls for: Add reviewed Unity Catalog table and column comments, owners and business definitions for consumer discovery.
Question 131
For this Azure Databricks issue: A Spark SQL pipeline spends most of its time executing wide joins and aggregations but the team has not evaluated a compatible vectorized execution engine. Engineers choose: Benchmark Photon on a supported runtime with representative queries and compare cost, latency and correctness. What acceptance evidence is BEST?
- A concurrent-user benchmark meets percentile latency and queue-depth targets with acceptable cost.
- Resource telemetry shows accelerator use where needed and CPU jobs satisfy their SLA at lower cost.
- Identical data and query workloads yield reconciled outputs and a recorded improvement in elapsed time per unit cost.
- A test upgrade on the interactive cluster does not change the isolated finance job's dependency versions or results.
Explanation
Correct Answer
C:The strongest acceptance proof is: Identical data and query workloads yield reconciled outputs and a recorded improvement in elapsed time per unit cost. It specifically verifies: Benchmark Photon on a supported runtime with representative queries and compare cost, latency and correctness. The original issue was: A Spark SQL pipeline spends most of its time executing wide joins and aggregations but the team has not evaluated a compatible vectorized execution engine. The underlying mechanism is: Photon can accelerate many supported SQL and DataFrame operations but requires workload-specific measurements. Without it, assuming every transformation accelerates equally may raise costs without reducing job duration.
Incorrect Answers
A:This result would validate a different measure: Tune supported SQL warehouse capacity and concurrency scaling based on simultaneous production query measurements. That is relevant to the distinct problem: A BI warehouse repeatedly slows during morning sign-in peaks even though a single analyst query runs quickly. Here the acceptance criterion is: Identical data and query workloads yield reconciled outputs and a recorded improvement in elapsed time per unit cost.
B:This result would validate a different measure: Assign GPU-capable compute only to supported GPU workloads and use CPU compute for ordinary transformations. That is relevant to the distinct problem: A feature-engineering notebook requires GPU libraries but the data cleaning jobs never use accelerators. Here the acceptance criterion is: Identical data and query workloads yield reconciled outputs and a recorded improvement in elapsed time per unit cost.
D:This result would validate a different measure: Run the finance job on dedicated supported job compute with its own pinned libraries and permissions. That is relevant to the distinct problem: A notebook pipeline uses a shared cluster with unstable third-party dependencies while a critical finance job needs controlled isolation. Here the acceptance criterion is: Identical data and query workloads yield reconciled outputs and a recorded improvement in elapsed time per unit cost.
Question 132
A data product owner adopts: Evaluate compatible instance pools or supported serverless startup behavior rather than keeping unused all-purpose clusters awake. The reason is: A scheduled workflow launches dozens of short-lived classic clusters and loses several minutes each time waiting for virtual machines. What observed result BEST proves effectiveness?
- Job run traces show lower startup latency and the combined pool or serverless cost stays within budget.
- A forced restart of interactive compute leaves the scheduled production task unaffected.
- A test upgrade on the interactive cluster does not change the isolated finance job's dependency versions or results.
- Spark UI and driver metrics show bounded driver memory after correcting the collection pattern.
Explanation
Correct Answer
A:The strongest acceptance proof is: Job run traces show lower startup latency and the combined pool or serverless cost stays within budget. It specifically verifies: Evaluate compatible instance pools or supported serverless startup behavior rather than keeping unused all-purpose clusters awake. The original issue was: A scheduled workflow launches dozens of short-lived classic clusters and loses several minutes each time waiting for virtual machines. The underlying mechanism is: Pool capacity can shorten classic cluster startup but reserved idle capacity must still be budgeted. Without it, ignoring startup overhead can make small jobs miss a narrow delivery window even when their Spark tasks are fast.
Incorrect Answers
B:This result would validate a different measure: Define a job-owned execution identity and eligible task compute separate from ad hoc analyst cluster lifecycle. That is relevant to the distinct problem: A scheduled ETL run uses shared analyst compute and fails whenever the cluster owner restarts a notebook session. Here the acceptance criterion is: Job run traces show lower startup latency and the combined pool or serverless cost stays within budget.
C:This result would validate a different measure: Run the finance job on dedicated supported job compute with its own pinned libraries and permissions. That is relevant to the distinct problem: A notebook pipeline uses a shared cluster with unstable third-party dependencies while a critical finance job needs controlled isolation. Here the acceptance criterion is: Job run traces show lower startup latency and the combined pool or serverless cost stays within budget.
D:This result would validate a different measure: Inspect driver collection and query planning behavior before simply enlarging every worker. That is relevant to the distinct problem: A SQL transformation consumes large amounts of memory on the driver despite executor tasks processing most source partitions. Here the acceptance criterion is: Job run traces show lower startup latency and the combined pool or serverless cost stays within budget.
Question 133
A data product owner adopts: Include performance and end-to-end tests using representative scale and data distributions. The reason is: An engineering test suite passes on miniature in-memory data yet fails in production on skewed billion-row joins. What observed result BEST proves effectiveness?
- A load test includes known hot keys and verifies latency, spills and output totals within targets.
- A staging run under the intended service principal reads and writes only its authorized objects.
- A rehearsed rollback restores a prior tested bundle and recovers the expected job definitions.
- An integration fixture with historical updates yields exactly the intended target versions.
Explanation
Correct Answer
A:The strongest acceptance proof is: A load test includes known hot keys and verifies latency, spills and output totals within targets. It specifically verifies: Include performance and end-to-end tests using representative scale and data distributions. The original issue was: An engineering test suite passes on miniature in-memory data yet fails in production on skewed billion-row joins. The underlying mechanism is: Correctness under small inputs does not establish workload performance at production cardinality. Without it, skew and memory limits may remain invisible until a full-scale run misses its SLA.
Incorrect Answers
B:This result would validate a different measure: Include integration tests of identity, Unity Catalog permissions and destination connectivity in staging. That is relevant to the distinct problem: A new pipeline passes unit tests but production fails because the service principal lacks a catalog grant. Here the acceptance criterion is: A load test includes known hot keys and verifies latency, spills and output totals within targets.
C:This result would validate a different measure: Tag and retain approved bundle releases and document a verified redeploy or rollback procedure. That is relevant to the distinct problem: An operations group wants to roll back a broken Lakeflow workflow but keeps no previous package version. Here the acceptance criterion is: A load test includes known hot keys and verifies latency, spills and output totals within targets.
D:This result would validate a different measure: Include an integration test using actual Delta table semantics and representative source changes. That is relevant to the distinct problem: A test harness mocks out Delta writes but the real job uses MERGE and unexpectedly changes historical rows. Here the acceptance criterion is: A load test includes known hot keys and verifies latency, spills and output totals within targets.
Question 134
The recommended measure is: Choose and document a supported format and catalog integration with explicit read/write compatibility tests. The operational context is: A cross-engine consumer expects a Delta table but source is plain Iceberg and engineers assume identical mutation semantics. Which data engineering concept applies?
- Fact grain determines which analyses can be reconstructed without information loss.
- CTAS populates a table at execution time and does not imply continuous synchronization with source.
- Delta version history can provide auditable historical snapshots while underlying files are retained.
- Delta and Iceberg are different transactional table formats with version-dependent feature support.
Explanation
Correct Answer
D:The decisive mechanism is: Delta and Iceberg are different transactional table formats with version-dependent feature support. That explains the selected action: Choose and document a supported format and catalog integration with explicit read/write compatibility tests. In the original scenario: A cross-engine consumer expects a Delta table but source is plain Iceberg and engineers assume identical mutation semantics. The specific unmanaged risk is: Assuming every engine can use every feature can result in invalid writes or unreadable metadata. Validate with: All required engines pass a controlled read and supported mutation matrix before production.
Incorrect Answers
A:This principle concerns a separate data-engineering condition: A facts table groups product sales at daily level although auditors must inspect individual transaction adjustments. Its targeted response is: Define fact-table grain at the required transaction or line-item level before aggregation. The current case is governed more directly by: Delta and Iceberg are different transactional table formats with version-dependent feature support.
B:This principle concerns a separate data-engineering condition: A pipeline uses CREATE TABLE AS SELECT for a one-time derived dataset but later assumes the result refreshes automatically. Its targeted response is: Select CTAS for an explicitly materialized snapshot or a scheduled pipeline for ongoing refresh. The current case is governed more directly by: Delta and Iceberg are different transactional table formats with version-dependent feature support.
C:This principle concerns a separate data-engineering condition: A business needs to compare two point-in-time table versions during a financial reconciliation. Its targeted response is: Use supported Delta table version history and time travel within the documented retention window. The current case is governed more directly by: Delta and Iceberg are different transactional table formats with version-dependent feature support.
Question 135
An implementation skips: Apply schema enforcement with explicit drift handling and alert on incompatible structural changes. Engineers are observing: A source API changes a quantity from integer to nested JSON and the pipeline silently produces nulls downstream. What consequence is MOST likely?
- Overly strict validation can block legitimate records while loose validation may permit missing keys.
- Invalid amounts may silently convert to null and skew revenue aggregates.
- A seemingly successful load can conceal a growing upstream data-loss problem.
- Silently coercing a breaking type can erase measurable values while a job appears successful.
Explanation
Correct Answer
D:The direct residual exposure is: Silently coercing a breaking type can erase measurable values while a job appears successful. The triggering condition is: A source API changes a quantity from integer to nested JSON and the pipeline silently produces nulls downstream. The omitted safeguard is: Apply schema enforcement with explicit drift handling and alert on incompatible structural changes. This matters because: Schema evolution must distinguish permissible additive fields from breaking type changes. A corrective test would show: A simulated type change is rejected or rescued with an actionable alert and no corrupted curated output.
Incorrect Answers
A:This failure mode belongs mainly to another situation: A new customer feed is missing optional contact information for 8% of rows and the team treats every null as pipeline failure. Its preventative action would be: Profile source missingness and define field-specific required versus optional expectations. The most direct exposure in the current scenario is: Silently coercing a breaking type can erase measurable values while a job appears successful.
B:This failure mode belongs mainly to another situation: A product data contract requires every transaction amount to use nonnegative decimal values but new raw files include quoted text strings. Its preventative action would be: Apply type, nullability and range expectations before promoting bronze rows into the silver target. The most direct exposure in the current scenario is: Silently coercing a breaking type can erase measurable values while a job appears successful.
C:This failure mode belongs mainly to another situation: A pipeline reports zero failed rows because bad records are silently dropped without any count of what was removed. Its preventative action would be: Collect expectation metrics and reconcile input, accepted, quarantined and rejected record counts. The most direct exposure in the current scenario is: Silently coercing a breaking type can erase measurable values while a job appears successful.
Question 136
The following control has NOT been implemented: Configure job failure notifications to approved recipients and validate incident routing. In this case: A scheduled Lakeflow Job has no owner alert when its final task fails overnight. Which exposure remains?
- A clock-only trigger can start before the dependency has produced the expected dataset.
- Mixed processing dates across tasks can create unreconciled reports and repeated records.
- An implicit UTC schedule can cause delivery an hour or more away from the required business cutover.
- A missed overnight job can remain unnoticed until business users complain.
Explanation
Correct Answer
D:The direct residual exposure is: A missed overnight job can remain unnoticed until business users complain. The triggering condition is: A scheduled Lakeflow Job has no owner alert when its final task fails overnight. The omitted safeguard is: Configure job failure notifications to approved recipients and validate incident routing. This matters because: Operational scheduling without actionable failure reporting is not sufficient for production reliability. A corrective test would show: A staged task failure produces an actionable notification with job ID and run link.
Incorrect Answers
A:This failure mode belongs mainly to another situation: A pipeline must run when an upstream Databricks table is updated but analysts have hardcoded a daily time estimate. Its preventative action would be: Evaluate supported table update triggers or a clear upstream job dependency based on actual availability. The most direct exposure in the current scenario is: A missed overnight job can remain unnoticed until business users complain.
B:This failure mode belongs mainly to another situation: A workflow needs to pass a date parameter from a scheduled trigger into all its downstream notebook tasks. Its preventative action would be: Define validated job parameters and propagate them through supported task parameter bindings. The most direct exposure in the current scenario is: A missed overnight job can remain unnoticed until business users complain.
C:This failure mode belongs mainly to another situation: A business report must refresh daily at 06:00 in a defined local time zone but a developer schedules it in UTC without conversion. Its preventative action would be: Configure the Lakeflow Job schedule with the intended time zone and verify daylight-saving behavior where relevant. The most direct exposure in the current scenario is: A missed overnight job can remain unnoticed until business users complain.
Question 137
For this Microsoft DP-750 case: A query on a Delta table slows after a small percentage of rows are deleted but data files are not optimized. Which design or operational response BEST fits?
- Review deletion-vector behavior, table layout and suitable optimization for the supported runtime.
- Inspect query predicates and Delta data skipping or clustering layout before increasing warehouse size again.
- Analyze concurrent query load and scaling policies rather than selecting capacity from daily averages.
- Define meaningful latency, error and backlog thresholds with alert routing based on the business SLO.
Explanation
Correct Answer
A:The most appropriate action is: Review deletion-vector behavior, table layout and suitable optimization for the supported runtime. It addresses the stated condition: A query on a Delta table slows after a small percentage of rows are deleted but data files are not optimized. The technical basis is: Logical deletes can leave physical layout and scan behavior needing maintenance. Without this control, ignoring metadata and file-layout changes can increase latency even though row counts decline. A direct acceptance check is: A supported optimization produces matching visible rows and a measured reduction in scanned data.
Incorrect Answers
B:This alternative primarily addresses a different Azure Databricks problem: A warehouse query remains slow even after larger compute is allocated and the profile reports scanning nearly every file. The corresponding measure there would be: Inspect query predicates and Delta data skipping or clustering layout before increasing warehouse size again. The stated situation instead calls for: Review deletion-vector behavior, table layout and suitable optimization for the supported runtime.
C:This alternative primarily addresses a different Azure Databricks problem: An interactive warehouse has high query queues during the morning but low CPU utilization after lunch. The corresponding measure there would be: Analyze concurrent query load and scaling policies rather than selecting capacity from daily averages. The stated situation instead calls for: Review deletion-vector behavior, table layout and suitable optimization for the supported runtime.
D:This alternative primarily addresses a different Azure Databricks problem: An alert triggers on every harmless microbatch delay while real hour-long ingestion stalls go unnoticed. The corresponding measure there would be: Define meaningful latency, error and backlog thresholds with alert routing based on the business SLO. The stated situation instead calls for: Review deletion-vector behavior, table layout and suitable optimization for the supported runtime.
Question 138
For this Azure Databricks issue: A batch enrichment task downloads a remote file and sometimes receives transient HTTP 503 responses. Engineers choose: Add bounded retries with exponential backoff and observable failure after the retry budget. What acceptance evidence is BEST?
- A deliberately failed write marks the job unsuccessful and prevents dependent publish tasks.
- A simulated final-task failure can be repaired without rerunning already validated upstream ingestion.
- An injected transient 503 recovers within the retry budget while repeated failures produce actionable alerts.
- A simulated hung request times out, alerts the owner and avoids publishing incomplete data.
Explanation
Correct Answer
C:The strongest acceptance proof is: An injected transient 503 recovers within the retry budget while repeated failures produce actionable alerts. It specifically verifies: Add bounded retries with exponential backoff and observable failure after the retry budget. The original issue was: A batch enrichment task downloads a remote file and sometimes receives transient HTTP 503 responses. The underlying mechanism is: Retryable transient failures should be distinguished from permanent data or permission errors. Without it, immediate permanent failure on a brief service outage can miss a delivery SLA.
Incorrect Answers
A:This result would validate a different measure: Propagate nonrecoverable exceptions to Lakeflow Jobs and attach actionable error metadata. That is relevant to the distinct problem: A workflow uses a notebook task that catches exceptions and returns success even when a write step fails. Here the acceptance criterion is: An injected transient 503 recovers within the retry budget while repeated failures produce actionable alerts.
B:This result would validate a different measure: Split ingestion, validation and publishing into explicit orchestrated tasks with durable intermediate outputs. That is relevant to the distinct problem: A multi-step notebook workflow handles source extraction and final reporting in one cell sequence with no recoverable checkpoints. Here the acceptance criterion is: An injected transient 503 recovers within the retry budget while repeated failures produce actionable alerts.
D:This result would validate a different measure: Use controlled task-level timeouts and isolate unpredictable external calls from bounded transform stages. That is relevant to the distinct problem: A scheduled notebook contains both transformation SQL and a long-running external API loop that occasionally hangs. Here the acceptance criterion is: An injected transient 503 recovers within the retry budget while repeated failures produce actionable alerts.
Question 139
For this Azure Databricks issue: An interactive warehouse has high query queues during the morning but low CPU utilization after lunch. Engineers choose: Analyze concurrent query load and scaling policies rather than selecting capacity from daily averages. What acceptance evidence is BEST?
- A skew-aware plan balances task durations and reduces the critical-stage elapsed time.
- A simulated failure yields archived logs, failed stage IDs and a documented root-cause path.
- A peak-concurrency replay meets latency and queue targets with recorded cost.
- Spark storage metrics show released cache and unchanged correctness with improved total runtime.
Explanation
Correct Answer
C:The strongest acceptance proof is: A peak-concurrency replay meets latency and queue targets with recorded cost. It specifically verifies: Analyze concurrent query load and scaling policies rather than selecting capacity from daily averages. The original issue was: An interactive warehouse has high query queues during the morning but low CPU utilization after lunch. The underlying mechanism is: Concurrent demand peaks can drive queue delays independently of average compute utilization. Without it, A warehouse sized to average load may breach dashboard p95 latency during business opening.
Incorrect Answers
A:This result would validate a different measure: Inspect the Spark UI DAG and skewed-key distribution before changing cluster size. That is relevant to the distinct problem: A production Spark query is slow and its stage graph shows one task running far longer than all peers. Here the acceptance criterion is: A peak-concurrency replay meets latency and queue targets with recorded cost.
B:This result would validate a different measure: Preserve diagnostic events and Spark UI stage evidence before restart where possible. That is relevant to the distinct problem: An operator restarts a failed Databricks cluster before collecting the Spark executor and driver logs. Here the acceptance criterion is: A peak-concurrency replay meets latency and queue targets with recorded cost.
D:This result would validate a different measure: Use caching only for beneficial reuse and release persisted state when no longer needed. That is relevant to the distinct problem: A team enables Spark cache for a one-time ETL dataset and never unpersists it before processing the next large table. Here the acceptance criterion is: A peak-concurrency replay meets latency and queue targets with recorded cost.
Question 140
The following control has NOT been implemented: Use appropriate table history, schema change records and approved retention/audit evidence rather than relying on memory. In this case: A shared table includes an old regulated field and the business needs reproducible evidence of when it was removed. Which exposure remains?
- Expired partners may retain ongoing data access despite a terminated agreement.
- Manual protection drift can expose fields on newly created tables before a review.
- Deleting a field without documented change history can leave unanswerable regulatory questions.
- Duplicated datasets can diverge or escape synchronized retention and correction policies.
Explanation
Correct Answer
C:The direct residual exposure is: Deleting a field without documented change history can leave unanswerable regulatory questions. The triggering condition is: A shared table includes an old regulated field and the business needs reproducible evidence of when it was removed. The omitted safeguard is: Use appropriate table history, schema change records and approved retention/audit evidence rather than relying on memory. This matters because: Metadata history and audit trails can support investigation of structure and access changes. A corrective test would show: The change record ties the removed column to an authorized deployment and preserved evidence.
Incorrect Answers
A:This failure mode belongs mainly to another situation: A Delta Sharing recipient must stop receiving future updates after an agreement expires. Its preventative action would be: Revoke or disable the relevant Delta Sharing recipient/share permissions and verify subsequent access. The most direct exposure in the current scenario is: Deleting a field without documented change history can leave unanswerable regulatory questions.
B:This failure mode belongs mainly to another situation: A central security team manually adds row masking rules to every department table and new datasets repeatedly lack protection. Its preventative action would be: Create compatible central ABAC policies based on governed classifications and verify inherited coverage. The most direct exposure in the current scenario is: Deleting a field without documented change history can leave unanswerable regulatory questions.
D:This failure mode belongs mainly to another situation: A shared customer table needs dynamic per-region row restrictions but the business refuses duplicating physical datasets. Its preventative action would be: Use row-filter policies on the governed table and test permitted region entitlements. The most direct exposure in the current scenario is: Deleting a field without documented change history can leave unanswerable regulatory questions.