Snowflake SnowPro Core COF-C03 Practice Test 53 - Prepaway
Practice Exams:

Pass Snowflake SnowPro Core COF-C03 Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers!

30 Days Free Updates, Instant Download!

Verified By Experts
Premium Bundle
$14.99
$74.99 $89.98

SnowPro Core COF-C03 Premium Bundle


  • Premium File 116 Questions & Answers. Last update: Nov 17, 2026
  • Study Guide 222 Pages

Snowflake SnowPro Core COF-C03 Practice Test 53

Question 520

Administrators at a logistics operator plan: Grant only necessary USAGE on parent database and schema plus table SELECT through an approved role. They are addressing the following problem in shipment events and distribution forecast tables: A team grants usage on a database but expects that grant to permit a reporting role to read all tables inside a schema. Which test demonstrates resolution?

  1. The service role performs the intended pipe operation but remains unable to administer unrelated objects.
  2. An authorized report reads the target table and cannot access an unrelated schema.
  3. A test execution completes the controlled multi-step process with an auditable outcome and no unsupported UDF side effects.
  4. The permitted operation succeeds under the correct role and an unrelated privileged action is denied.

Correct Answer: B

Explanation

A.

The service role performs the intended pipe operation but remains unable to administer unrelated objects. Pipes, stages, file formats, and tables are distinct Snowflake objects with different privileges. For shipment events and distribution forecast tables at a logistics operator, the role can resolve the database name but lacks schema usage and SELECT privileges on the required table. Verify that an authorized report reads the target table and cannot access an unrelated schema.

B.

An authorized report reads the target table and cannot access an unrelated schema. For shipment events and distribution forecast tables at a logistics operator, the role can resolve the database name but lacks schema usage and SELECT privileges on the required table. Object visibility and data access depend on privileges at multiple securable-object hierarchy levels. To address this, grant only necessary USAGE on parent database and schema plus table SELECT through an approved role.

C.

A test execution completes the controlled multi-step process with an auditable outcome and no unsupported UDF side effects. Snowflake functions and procedures have different execution contracts and appropriate use cases. For shipment events and distribution forecast tables at a logistics operator, the role can resolve the database name but lacks schema usage and SELECT privileges on the required table. Verify that an authorized report reads the target table and cannot access an unrelated schema.

D.

The permitted operation succeeds under the correct role and an unrelated privileged action is denied. Snowsight executes SQL within the selected session role, database, schema and warehouse context. For shipment events and distribution forecast tables at a logistics operator, the role can resolve the database name but lacks schema usage and SELECT privileges on the required table. Verify that an authorized report reads the target table and cannot access an unrelated schema.

Question 521

A Snowflake analyst supporting cargo scans and berth utilization reports at a container port operator intends: Review primary and secondary active-role authorization and remove excess grants from the underlying responsible roles. The reported condition is: A role investigation finds a user can access a table even after the primary active role lost its SELECT privilege. Identify the most decisive diagnostic record.

  1. Role hierarchy records show the database role is not granted through an account role with supported activation.
  2. GRANTS inventory shows table SELECT is present but warehouse USAGE is missing from active role hierarchy.
  3. A low-privilege test session selects cleartext identifiers because no applicable masking policy protects the column.
  4. Session diagnostics show an enabled secondary role still grants access to the table.

Correct Answer: D

Explanation

A.

Role hierarchy records show the database role is not granted through an account role with supported activation. This choice does not explain the observed failure mechanism. For cargo scans and berth utilization reports at a container port operator, session diagnostics show an enabled secondary role still grants access to the table. To address this, review primary and secondary active-role authorization and remove excess grants from the underlying responsible roles.

B.

GRANTS inventory shows table SELECT is present but warehouse USAGE is missing from active role hierarchy. This choice does not explain the observed failure mechanism. For cargo scans and berth utilization reports at a container port operator, session diagnostics show an enabled secondary role still grants access to the table. To address this, review primary and secondary active-role authorization and remove excess grants from the underlying responsible roles.

C.

A low-privilege test session selects cleartext identifiers because no applicable masking policy protects the column. This choice does not explain the observed failure mechanism. For cargo scans and berth utilization reports at a container port operator, session diagnostics show an enabled secondary role still grants access to the table. To address this, review primary and secondary active-role authorization and remove excess grants from the underlying responsible roles.

D.

Session diagnostics show an enabled secondary role still grants access to the table. Snowflake authorization can aggregate primary and enabled secondary role privileges. The corresponding fix is to review primary and secondary active-role authorization and remove excess grants from the underlying responsible roles. A successful test should show that the user loses unauthorized read access in all permitted role combinations while approved data remains accessible.

Question 522

For assembly sensor events and warranty reports at an automotive parts manufacturer, Snowflake administrators supporting assembly sensor events and warranty reports report: A data provider publishes a marketplace listing publicly while the intended audience is a small contracted pilot group. Their diagnostic record states: Listing visibility and consumer entitlement inspection show unrestricted discovery beyond pilot participants. Which corrective action is best supported by the evidence?

  1. Choose an appropriate private listing and governed access workflow before publishing the data product.
  2. Review the Native App package, application roles, permission requests and version lifecycle with the consumer.
  3. Configure a supported secure data share with least privilege and document relevant region/account constraints.
  4. Use supported clean room constraints and approved collaboration templates to restrict outputs and data access.

Correct Answer: A

Explanation

A.

Choose an appropriate private listing and governed access workflow before publishing the data product. For assembly sensor events and warranty reports at an automotive parts manufacturer, listing visibility and consumer entitlement inspection show unrestricted discovery beyond pilot participants. Marketplace listing audience and share-grant design are independent governance controls. A suitable check should show that only authorized pilot consumers can discover and access the test listing while other accounts cannot.

B.

Review the Native App package, application roles, permission requests and version lifecycle with the consumer. Native Apps distribute governed in-platform logic, but installation privileges and lifecycle must still be reviewed. For assembly sensor events and warranty reports at an automotive parts manufacturer, listing visibility and consumer entitlement inspection show unrestricted discovery beyond pilot participants. The better-matched action is to choose an appropriate private listing and governed access workflow before publishing the data product.

C.

Configure a supported secure data share with least privilege and document relevant region/account constraints. Snowflake secure sharing can provide governed access without traditional physical data copying. For assembly sensor events and warranty reports at an automotive parts manufacturer, listing visibility and consumer entitlement inspection show unrestricted discovery beyond pilot participants. The better-matched action is to choose an appropriate private listing and governed access workflow before publishing the data product.

D.

Use supported clean room constraints and approved collaboration templates to restrict outputs and data access. Clean rooms can support collaboration on sensitive data under controlled computations and output rules. For assembly sensor events and warranty reports at an automotive parts manufacturer, listing visibility and consumer entitlement inspection show unrestricted discovery beyond pilot participants. The better-matched action is to choose an appropriate private listing and governed access workflow before publishing the data product.

Question 523

Administrators at a cooperative credit union plan: Benchmark cache eligibility using consistent authorized roles, parameter settings and unchanged data without exposing protected results. They are addressing the following problem in loan servicing analytics and audit trails: Two analysts compare query cache savings using different roles and assume an exact result cache hit is universally shareable. Which test demonstrates resolution?

  1. An unchanged eligible repeated query demonstrates the expected cache behavior without stale results.
  2. A repeat benchmark documents expected cold-start cost and maintains the agreed total workload SLA.
  3. A synthetic repeated key resolves to one correct target state and replay remains consistent.
  4. Repeated qualified queries show understood result-cache behavior with authorization preserved.

Correct Answer: D

Explanation

A.

An unchanged eligible repeated query demonstrates the expected cache behavior without stale results. Snowflake persisted query results are reused only when the required consistency and context conditions are met. For loan servicing analytics and audit trails at a cooperative credit union, query contexts differ in security and session conditions and tests do not control source change timing. A confirming test shows that repeated qualified queries show understood result-cache behavior with authorization preserved.

B.

A repeat benchmark documents expected cold-start cost and maintains the agreed total workload SLA. Virtual warehouse local data cache behavior differs from the persisted query result cache. For loan servicing analytics and audit trails at a cooperative credit union, query contexts differ in security and session conditions and tests do not control source change timing. A confirming test shows that repeated qualified queries show understood result-cache behavior with authorization preserved.

C.

A synthetic repeated key resolves to one correct target state and replay remains consistent. MERGE correctness requires an unambiguous source-to-target matching strategy. For loan servicing analytics and audit trails at a cooperative credit union, query contexts differ in security and session conditions and tests do not control source change timing. A confirming test shows that repeated qualified queries show understood result-cache behavior with authorization preserved.

D.

Repeated qualified queries show understood result-cache behavior with authorization preserved. For loan servicing analytics and audit trails at a cooperative credit union, query contexts differ in security and session conditions and tests do not control source change timing. Query result reuse depends on relevant session and security conditions as well as SQL text. To correct it, benchmark cache eligibility using consistent authorized roles, parameter settings and unchanged data without exposing protected results.

Question 524

A Snowflake analyst supporting fleet schedules and passenger ridership datasets at a regional transit agency intends: Introduce governed query tags and accountable job identity mapping with account usage reporting. The reported condition is: A FinOps team sees high daily query spend but cannot attribute it to the dashboard or scheduled transform responsible. Identify the most decisive diagnostic record.

  1. Query History separates long queued time from short actual execution and shows many simultaneous sessions.
  2. Query Profile shows high local and remote spill even though the scan filters relevant partitions correctly.
  3. Query history shows new execution work and source-table modification timestamps between otherwise similar report runs.
  4. Query history contains query IDs but service account and tagging conventions do not map runs to owning workloads.

Correct Answer: D

Explanation

A.

Query History separates long queued time from short actual execution and shows many simultaneous sessions. This choice does not explain the observed failure mechanism. For fleet schedules and passenger ridership datasets at a regional transit agency, query history contains query IDs but service account and tagging conventions do not map runs to owning workloads. Instead, introduce governed query tags and accountable job identity mapping with account usage reporting.

B.

Query Profile shows high local and remote spill even though the scan filters relevant partitions correctly. This choice does not explain the observed failure mechanism. For fleet schedules and passenger ridership datasets at a regional transit agency, query history contains query IDs but service account and tagging conventions do not map runs to owning workloads. Instead, introduce governed query tags and accountable job identity mapping with account usage reporting.

C.

Query history shows new execution work and source-table modification timestamps between otherwise similar report runs. This choice does not explain the observed failure mechanism. For fleet schedules and passenger ridership datasets at a regional transit agency, query history contains query IDs but service account and tagging conventions do not map runs to owning workloads. Instead, introduce governed query tags and accountable job identity mapping with account usage reporting.

D.

Query history contains query IDs but service account and tagging conventions do not map runs to owning workloads. Query optimization governance needs observable workload attribution, not only raw credits. To address this, introduce governed query tags and accountable job identity mapping with account usage reporting. Confirm that a seeded expensive query is assigned to the correct operational owner in cost reporting; the reported setting is: a snowflake analyst supporting fleet schedules and passenger ridership datasets at a regional transit agency intends: introduce governed query tags and accountable job identity mapping with account usage r.

Question 525

For care utilization metrics and de-identified clinical reporting at a hospital network, Snowflake administrators supporting care utilization metrics and de-identified clinical reporting report: A monthly report scans the full sales history after its date predicate was rewritten as a formatted string expression. Their diagnostic record states: Query Profile shows much higher scanned micro-partition counts after the change with identical date filter intent. Which corrective action is best supported by the evidence?

  1. Restore a selective supported date-range predicate and compare query profile pruning under reproducible data.
  2. Review join/sort memory needs, SQL shape and approved warehouse capacity before changing clustering keys.
  3. Correct join cardinality or pre-aggregate event data at the required grain before publishing financial metrics.
  4. Choose an optimization suited to large aggregate access, such as SQL restructuring or appropriate precomputation, after benchmarking.

Correct Answer: A

Explanation

A.

Restore a selective supported date-range predicate and compare query profile pruning under reproducible data. For care utilization metrics and de-identified clinical reporting at a hospital network, query Profile shows much higher scanned micro-partition counts after the change with identical date filter intent. Pruning efficiency depends on how filter predicates interact with micro-partition metadata. The expected outcome is that same expected rows return with fewer partitions scanned and lower measured query cost.

B.

Review join/sort memory needs, SQL shape and approved warehouse capacity before changing clustering keys. Query Profile distinguishes memory spill from excessive table scans and pruning inefficiency. For care utilization metrics and de-identified clinical reporting at a hospital network, query Profile shows much higher scanned micro-partition counts after the change with identical date filter intent. To address this, restore a selective supported date-range predicate and compare query profile pruning under reproducible data.

C.

Correct join cardinality or pre-aggregate event data at the required grain before publishing financial metrics. Join cardinality affects both query cost and semantic result correctness. For care utilization metrics and de-identified clinical reporting at a hospital network, query Profile shows much higher scanned micro-partition counts after the change with identical date filter intent. To address this, restore a selective supported date-range predicate and compare query profile pruning under reproducible data.

D.

Choose an optimization suited to large aggregate access, such as SQL restructuring or appropriate precomputation, after benchmarking. Search optimization is designed for selective access patterns, not all-table analytic scans. For care utilization metrics and de-identified clinical reporting at a hospital network, query Profile shows much higher scanned micro-partition counts after the change with identical date filter intent. To address this, restore a selective supported date-range predicate and compare query profile pruning under reproducible data.

Question 526

At an e-commerce marketplace, a Snowflake analyst supporting shopping funnel events and merchant sales reporting finds: A marketing dataset stores several item entries in a VARIANT array but a report treats the entire array as one item. They confirm: A sample event contains multiple nested elements and the output row count stays one per parent regardless of item count. Select the action and acceptance-test pair best matched to these facts.

  1. Use a supported tolerant conversion and explicit data-quality exception handling while retaining the original VARIANT. — verification: Valid numeric fields transform correctly and missing values are flagged without silently inventing numbers.
  2. Use the appropriate LATERAL FLATTEN and reviewed join grain to expand the intended nested elements. — verification: A three-element test array produces three authorized item rows with correct parent identifiers.
  3. Use the correct business key in PARTITION BY with explicit deterministic ORDER BY where needed. — verification: Each test customer has the expected independent ranking without unexpected cross-customer influence.
  4. Correct join cardinality or pre-aggregate event data at the required grain before publishing financial metrics. — verification: A seeded order contributes its intended total once and query row expansion returns to expected levels.

Correct Answer: B

Explanation

A.

Use a supported tolerant conversion and explicit data-quality exception handling while retaining the original VARIANT. Verify that valid numeric fields transform correctly and missing values are flagged without silently inventing numbers. Semi-structured data can contain absent or variant-typed attributes requiring explicit conversion behavior. For shopping funnel events and merchant sales reporting at an e-commerce marketplace, a sample event contains multiple nested elements and the output row count stays one per parent regardless of item count. The appropriate correction is to use the appropriate LATERAL FLATTEN and reviewed join grain to expand the intended nested elements. Confirm that a three-element test array produces three authorized item rows with correct parent identifiers.

B.

Use the appropriate LATERAL FLATTEN and reviewed join grain to expand the intended nested elements. Verify that a three-element test array produces three authorized item rows with correct parent identifiers. For shopping funnel events and merchant sales reporting at an e-commerce marketplace, a sample event contains multiple nested elements and the output row count stays one per parent regardless of item count. Flattening nested semi-structured arrays changes row granularity and requires cardinality awareness.

C.

Use the correct business key in PARTITION BY with explicit deterministic ORDER BY where needed. Verify that Each test customer has the expected independent ranking without unexpected cross-customer influence. Window functions depend on the declared partition and ordering semantics. For shopping funnel events and merchant sales reporting at an e-commerce marketplace, a sample event contains multiple nested elements and the output row count stays one per parent regardless of item count. The appropriate correction is to use the appropriate LATERAL FLATTEN and reviewed join grain to expand the intended nested elements. Confirm that a three-element test array produces three authorized item rows with correct parent identifiers.

D.

Correct join cardinality or pre-aggregate event data at the required grain before publishing financial metrics. Verify that a seeded order contributes its intended total once and query row expansion returns to expected levels. Join cardinality affects both query cost and semantic result correctness. For shopping funnel events and merchant sales reporting at an e-commerce marketplace, a sample event contains multiple nested elements and the output row count stays one per parent regardless of item count. The appropriate correction is to use the appropriate LATERAL FLATTEN and reviewed join grain to expand the intended nested elements. Confirm that a three-element test array produces three authorized item rows with correct parent identifiers.

Question 527

The fleet schedules and passenger ridership datasets team at a regional transit agency records: A developer runs the same SQL from an IDE and Snowsight but receives a table-not-found error only in the IDE. Logs confirm: The IDE connection has a different current database and schema and unqualified table names resolve to a stale namespace. The recommended action is: Set explicit intended database and schema context or fully qualify production-sensitive object references. What underlying Snowflake solution principle applies?

  1. Interactive development tools still require supported execution compute and role context.
  2. Snowsight executes SQL within the selected session role, database, schema and warehouse context.
  3. Temporary tables are scoped to sessions and should not be treated as persistent shared datasets.
  4. Snowflake session context and object qualification determine how SQL names are resolved.

Correct Answer: D

Explanation

A.

Interactive development tools still require supported execution compute and role context. This choice does not explain the observed failure mechanism. For fleet schedules and passenger ridership datasets at a regional transit agency, the IDE connection has a different current database and schema and unqualified table names resolve to a stale namespace. The controlling principle is that Snowflake session context and object qualification determine how SQL names are resolved.

B.

Snowsight executes SQL within the selected session role, database, schema and warehouse context. This choice does not explain the observed failure mechanism. For fleet schedules and passenger ridership datasets at a regional transit agency, the IDE connection has a different current database and schema and unqualified table names resolve to a stale namespace. The controlling principle is that Snowflake session context and object qualification determine how SQL names are resolved.

C.

Temporary tables are scoped to sessions and should not be treated as persistent shared datasets. This choice does not explain the observed failure mechanism. For fleet schedules and passenger ridership datasets at a regional transit agency, the IDE connection has a different current database and schema and unqualified table names resolve to a stale namespace. The controlling principle is that Snowflake session context and object qualification determine how SQL names are resolved.

D.

Snowflake session context and object qualification determine how SQL names are resolved. For fleet schedules and passenger ridership datasets at a regional transit agency, the IDE connection has a different current database and schema and unqualified table names resolve to a stale namespace. To correct it, set explicit intended database and schema context or fully qualify production-sensitive object references. A successful test shows both approved client sessions resolve the same intended table without using elevated roles.

Question 528

The Snowflake data engineering team at a public university investigating student analytics and research-grant reporting notes: A finance warehouse auto-suspends overnight but scheduled morning jobs fail because the service cannot resume it. Evidence: Warehouse state is suspended, the scheduling role has usage permissions, and auto-resume is disabled for the chosen warehouse. They plan: Review warehouse auto-resume and approved execution role settings without disabling cost-saving auto-suspend entirely. Which technical principle justifies the correction?

  1. Multi-cluster warehouses address concurrency pressure differently from single-query compute capacity.
  2. Warehouse auto-suspend and auto-resume must be configured together for intermittent workloads.
  3. Warehouse scale-up and multi-cluster scale-out address different performance bottlenecks.
  4. Interactive development tools still require supported execution compute and role context.

Correct Answer: B

Explanation

A.

Multi-cluster warehouses address concurrency pressure differently from single-query compute capacity. This choice does not explain the observed failure mechanism. For student analytics and research-grant reporting at a public university, warehouse state is suspended, the scheduling role has usage permissions, and auto-resume is disabled for the chosen warehouse. The controlling principle is that warehouse auto-suspend and auto-resume must be configured together for intermittent workloads.

B.

Warehouse auto-suspend and auto-resume must be configured together for intermittent workloads. For student analytics and research-grant reporting at a public university, warehouse state is suspended, the scheduling role has usage permissions, and auto-resume is disabled for the chosen warehouse. To address this, review warehouse auto-resume and approved execution role settings without disabling cost-saving auto-suspend entirely. A successful test shows a scheduled test query resumes the warehouse, succeeds and the warehouse later suspends again after inactivity.

C.

Warehouse scale-up and multi-cluster scale-out address different performance bottlenecks. This choice does not explain the observed failure mechanism. For student analytics and research-grant reporting at a public university, warehouse state is suspended, the scheduling role has usage permissions, and auto-resume is disabled for the chosen warehouse. The controlling principle is that warehouse auto-suspend and auto-resume must be configured together for intermittent workloads.

D.

Interactive development tools still require supported execution compute and role context. This choice does not explain the observed failure mechanism. For student analytics and research-grant reporting at a public university, warehouse state is suspended, the scheduling role has usage permissions, and auto-resume is disabled for the chosen warehouse. The controlling principle is that warehouse auto-suspend and auto-resume must be configured together for intermittent workloads.

Question 529

During a change review for building event streams and supplier invoices at a facilities management group, the team finds: An account team promises an edition-dependent governance control to auditors without checking the account's subscribed Snowflake edition. Investigation shows: Feature evaluation indicates the required control is not available at the current account edition and its trial behavior differs from production. What is the most appropriate next Snowflake action?

  1. Separate warehouse compute lifecycle from durable database storage and enable supported auto-suspend with a safe restart test.
  2. Verify official feature and edition availability and select an approved edition or alternative control before making compliance claims.
  3. Inspect query planning, object statistics, metadata and query shape before changing compute sizing.
  4. Use explicit vetted CLI connection profiles, validate active account and role, and add preflight context assertions to deployment.

Correct Answer: B

Explanation

A.

Separate warehouse compute lifecycle from durable database storage and enable supported auto-suspend with a safe restart test. Snowflake separates independently scalable query compute from persistent managed table storage. For building event streams and supplier invoices at a facilities management group, feature evaluation indicates the required control is not available at the current account edition and its trial behavior differs from production. To address this, verify official feature and edition availability and select an approved edition or alternative control before making compliance claims.

B.

Verify official feature and edition availability and select an approved edition or alternative control before making compliance claims. For building event streams and supplier invoices at a facilities management group, feature evaluation indicates the required control is not available at the current account edition and its trial behavior differs from production. Snowflake editions differ in available advanced security and resilience capabilities. The expected outcome is that a controlled account check demonstrates that the chosen production configuration actually enforces the documented control.

C.

Inspect query planning, object statistics, metadata and query shape before changing compute sizing. Cloud services coordinate SQL compilation and metadata separately from virtual warehouse execution. For building event streams and supplier invoices at a facilities management group, feature evaluation indicates the required control is not available at the current account edition and its trial behavior differs from production. To address this, verify official feature and edition availability and select an approved edition or alternative control before making compliance claims.

D.

Use explicit vetted CLI connection profiles, validate active account and role, and add preflight context assertions to deployment. A successful client connection does not establish that the selected account context is correct. For building event streams and supplier invoices at a facilities management group, feature evaluation indicates the required control is not available at the current account edition and its trial behavior differs from production. To address this, verify official feature and edition availability and select an approved edition or alternative control before making compliance claims.

Question 530

In booking data marts and occupancy forecasts at a hospitality chain, a validated correction is proposed: Correct and test a documented row access policy driven by authoritative entitlement mapping without using a broad global exception. It responds to: A shared case table returns records for customers in unrelated business units under a narrowly scoped analyst role. Choose the best measurable success criterion.

  1. Restricted roles receive redacted test values and a specifically authorized role receives only approved detail.
  2. Positive and negative tenant tests return only authorized records for each role.
  3. A seeded new sensitive column is detected, labeled appropriately and included in the audit inventory.
  4. The user's supported active account role inherits authorized database object privileges but no unrelated account-wide access.

Correct Answer: B

Explanation

A.

Restricted roles receive redacted test values and a specifically authorized role receives only approved detail. This choice does not explain the observed failure mechanism. For booking data marts and occupancy forecasts at a hospitality chain, a controlled session can query another unit's case rows because row access policy predicates do not use the correct tenant mapping. The outcome should establish that positive and negative tenant tests return only authorized records for each role.

B.

Positive and negative tenant tests return only authorized records for each role. For booking data marts and occupancy forecasts at a hospitality chain, a controlled session can query another unit's case rows because row access policy predicates do not use the correct tenant mapping. Row-level security requires correct identity-to-row entitlement logic. The documented change is to correct and test a documented row access policy driven by authoritative entitlement mapping without using a broad global exception.

C.

A seeded new sensitive column is detected, labeled appropriately and included in the audit inventory. This choice does not explain the observed failure mechanism. For booking data marts and occupancy forecasts at a hospitality chain, a controlled session can query another unit's case rows because row access policy predicates do not use the correct tenant mapping. The outcome should establish that positive and negative tenant tests return only authorized records for each role.

D.

The user's supported active account role inherits authorized database object privileges but no unrelated account-wide access. This choice does not explain the observed failure mechanism. For booking data marts and occupancy forecasts at a hospitality chain, a controlled session can query another unit's case rows because row access policy predicates do not use the correct tenant mapping. The outcome should establish that positive and negative tenant tests return only authorized records for each role.

SnowPro Core COF-C03 Practice Tests