Question 270
A data engineer wants notebooks combining SQL and Python experiments while retaining controlled Snowflake data access. What is the most appropriate next design or operational decision?
- Inspect the session's active database, schema, role, and warehouse context before changing SQL.
- Use Snowflake Notebooks for supported iterative SQL and Python analysis.
- Use the Snowflake CLI for supported command-line project and deployment operations.
- Explain that Snowflake stores persisted table data in a storage layer separate from virtual warehouse compute.
Explanation
Correct Answer
B:Notebooks combine code, results, and explanatory cells within a Snowflake-integrated workspace. Success should be confirmed by this scenario-specific evidence: the engineer can rerun the notebook against authorized tables and inspect generated results.
Incorrect Answers
A:Unqualified object names and available privileges are interpreted using active session context. In contrast, notebooks combine code, results, and explanatory cells within a Snowflake-integrated workspace.
C:Snowflake CLI provides scriptable commands and project workflows suitable for automated execution. In contrast, notebooks combine code, results, and explanatory cells within a Snowflake-integrated workspace.
D:Independent storage and compute let a query warehouse change size while the same durable table data remains centrally managed. In contrast, notebooks combine code, results, and explanatory cells within a Snowflake-integrated workspace.
Question 271
A finance analyst sees cloud services charges and assumes every Snowflake operation bills solely through active warehouses. The team proposes this action: Review warehouse, cloud-services, serverless, and storage usage components separately. Why should that action address the stated requirement?
- Replication maintains supported copies for continuity, and failover procedures can redirect supported workloads.
- Database roles encapsulate database-scoped privileges and are inherited through granted account roles.
- Snowflake charges can originate from multiple metered services beyond user-managed warehouse compute.
- Centralized identity and additional authentication factors strengthen sign-in without replacing authorization.
Explanation
Correct Answer
C:Snowflake charges can originate from multiple metered services beyond user-managed warehouse compute. This explains the proposed action: review warehouse, cloud-services, serverless, and storage usage components separately.
Incorrect Answers
A:Replication maintains supported copies for continuity, and failover procedures can redirect supported workloads. That rationale explains plan Snowflake replication and failover capabilities with documented recovery requirements. It does not establish the mechanism needed here: Snowflake charges can originate from multiple metered services beyond user-managed warehouse compute.
B:Database roles encapsulate database-scoped privileges and are inherited through granted account roles. That rationale explains create a database role and grant it to an appropriate account role. It does not establish the mechanism needed here: Snowflake charges can originate from multiple metered services beyond user-managed warehouse compute.
D:Centralized identity and additional authentication factors strengthen sign-in without replacing authorization. That rationale explains configure supported federated authentication or SSO and enforce MFA according to account policy. It does not establish the mechanism needed here: Snowflake charges can originate from multiple metered services beyond user-managed warehouse compute.
Question 272
A security lead wants an account-wide view of relevant posture findings and recommended hardening checks. What is the most appropriate next design or operational decision?
- Use a managed access schema and perform grants through its owner or a role with MANAGE GRANTS.
- Configure and test an appropriate Snowflake network policy at the required scope.
- Use the object owner's discretionary grant authority, subject to Snowflake privilege rules.
- Review Snowflake Trust Center findings and appropriate security recommendations.
Explanation
Correct Answer
D:Trust Center helps surface posture information that complements focused policy configuration. Success should be confirmed by this scenario-specific evidence: the security team can triage findings and assign remediation owners.
Incorrect Answers
A:In managed access schemas, grant authority is centralized even when individual roles own tables. In contrast, trust Center helps surface posture information that complements focused policy configuration.
B:Network policies constrain the permitted network locations for connecting clients. In contrast, trust Center helps surface posture information that complements focused policy configuration.
C:Discretionary access control lets an object's owner grant allowed privileges in a regular schema. In contrast, trust Center helps surface posture information that complements focused policy configuration.
Question 273
A business analyst needs to inspect query history and open a worksheet without setting up local client software. The proposed response is: Use Snowsight to explore objects, run worksheets, and examine query history in the browser. Which limitation or trade-off should the team acknowledge?
- Snowsight does not grant data privileges that the active Snowflake role lacks.
- Cloud services are not a replacement for warehouse compute on ordinary table scans and joins.
- Changing warehouse type should be benchmarked; it does not universally improve SQL query speed.
- Automatic clustering has maintenance cost and is not justified for every small table.
Explanation
Correct Answer
A:Snowsight does not grant data privileges that the active Snowflake role lacks. It qualifies the proposed action: use Snowsight to explore objects, run worksheets, and examine query history in the browser.
Incorrect Answers
B:Cloud services are not a replacement for warehouse compute on ordinary table scans and joins. This is a trade-off for the separate proposal: identify the cloud services layer as the layer coordinating authentication, metadata, and query optimization. The proposed choice instead requires acknowledging this risk: Snowsight does not grant data privileges that the active Snowflake role lacks.
C:Changing warehouse type should be benchmarked; it does not universally improve SQL query speed. This is a trade-off for the separate proposal: assess a Snowpark-optimized warehouse type for the supported memory-intensive Snowpark workload. The proposed choice instead requires acknowledging this risk: Snowsight does not grant data privileges that the active Snowflake role lacks.
D:Automatic clustering has maintenance cost and is not justified for every small table. This is a trade-off for the separate proposal: evaluate a clustering key aligned with the common date filter and monitor reclustering cost. The proposed choice instead requires acknowledging this risk: Snowsight does not grant data privileges that the active Snowflake role lacks.
Question 274
An administrator needs historical query patterns by user, warehouse, and error state over the last reporting period. After implementing the approach, which observation best demonstrates the required outcome?
- The subsequent query reflects current committed data rather than an invalid stale result.
- A report isolates repeated failures or slow statements by the relevant owners and warehouses.
- Representative lookup queries show lower scans or latency after the optimization is ready.
- Warmup behavior explains the morning scan slowdown without implying data loss.
Explanation
Correct Answer
B:A report isolates repeated failures or slow statements by the relevant owners and warehouses. This demonstrates the outcome for the stated requirement rather than just a saved configuration.
Incorrect Answers
A:The subsequent query reflects current committed data rather than an invalid stale result. That measures a different outcome: A BI service frequently sees cached responses but an analyst modifies a source table between requests. In this case, verify that a report isolates repeated failures or slow statements by the relevant owners and warehouses.
C:Representative lookup queries show lower scans or latency after the optimization is ready. That measures a different outcome: a huge event table receives repeated point-lookups on a highly selective identifier. In this case, verify that a report isolates repeated failures or slow statements by the relevant owners and warehouses.
D:Warmup behavior explains the morning scan slowdown without implying data loss. That measures a different outcome: a finance team suspends its warehouse nightly and expects identical warm data-cache performance the next morning. In this case, verify that a report isolates repeated failures or slow statements by the relevant owners and warehouses.
Question 275
A database administrator wants reusable privileges scoped to one application database, but users activate account roles. The team proposes this action: Create a database role and grant it to an appropriate account role. Why should that action address the stated requirement?
- Object tagging adds governance metadata that can be queried and used in supported policy workflows.
- Trust Center helps surface posture information that complements focused policy configuration.
- Database roles encapsulate database-scoped privileges and are inherited through granted account roles.
- Snowflake charges can originate from multiple metered services beyond user-managed warehouse compute.
Explanation
Correct Answer
C:Database roles encapsulate database-scoped privileges and are inherited through granted account roles. This explains the proposed action: create a database role and grant it to an appropriate account role.
Incorrect Answers
A:Object tagging adds governance metadata that can be queried and used in supported policy workflows. That rationale explains use object tags and a controlled tagging process to annotate the intended data assets. It does not establish the mechanism needed here: database roles encapsulate database-scoped privileges and are inherited through granted account roles.
B:Trust Center helps surface posture information that complements focused policy configuration. That rationale explains review Snowflake Trust Center findings and appropriate security recommendations. It does not establish the mechanism needed here: database roles encapsulate database-scoped privileges and are inherited through granted account roles.
D:Snowflake charges can originate from multiple metered services beyond user-managed warehouse compute. That rationale explains review warehouse, cloud-services, serverless, and storage usage components separately. It does not establish the mechanism needed here: database roles encapsulate database-scoped privileges and are inherited through granted account roles.
Question 276
A Python data-processing job has per-worker memory pressure that ordinary warehouse tuning has not resolved. The proposed response is: Assess a Snowpark-optimized warehouse type for the supported memory-intensive Snowpark workload. Which limitation or trade-off should the team acknowledge?
- Creating a stage does not load the files into a target table by itself.
- Separate accounts do not by themselves synchronize data or establish cross-region failover.
- An Iceberg table is not simply an ordinary Snowflake-managed permanent table in all lifecycle respects.
- Changing warehouse type should be benchmarked; it does not universally improve SQL query speed.
Explanation
Correct Answer
D:Changing warehouse type should be benchmarked; it does not universally improve SQL query speed. It qualifies the proposed action: assess a Snowpark-optimized warehouse type for the supported memory-intensive Snowpark workload.
Incorrect Answers
A:Creating a stage does not load the files into a target table by itself. This is a trade-off for the separate proposal: create a named external stage referencing the authorized cloud storage location and credentials integration. The proposed choice instead requires acknowledging this risk: changing warehouse type should be benchmarked; it does not universally improve SQL query speed.
B:Separate accounts do not by themselves synchronize data or establish cross-region failover. This is a trade-off for the separate proposal: use separate Snowflake accounts for independent administration, warehouses, and account-level policy scopes. The proposed choice instead requires acknowledging this risk: changing warehouse type should be benchmarked; it does not universally improve SQL query speed.
C:An Iceberg table is not simply an ordinary Snowflake-managed permanent table in all lifecycle respects. This is a trade-off for the separate proposal: consider a supported Apache Iceberg table using the appropriate catalog and external storage configuration. The proposed choice instead requires acknowledging this risk: changing warehouse type should be benchmarked; it does not universally improve SQL query speed.
Question 277
A BI service frequently sees cached responses but an analyst modifies a source table between requests. The team proposes this action: Retest result cache eligibility after the source-data change instead of assuming reuse. Why should that action address the stated requirement?
- Snowflake does not promise reuse of a persisted result after the underlying result-relevant data changes.
- Separate warehouses provide compute isolation and independent concurrency controls over shared data.
- Poor partition pruning forces more data scanning even when returned row count is small.
- Aggregate correctness depends on matching the intended grain, not merely using SUM after an exploding join.
Explanation
Correct Answer
A:Snowflake does not promise reuse of a persisted result after the underlying result-relevant data changes. This explains the proposed action: retest result cache eligibility after the source-data change instead of assuming reuse.
Incorrect Answers
B:Separate warehouses provide compute isolation and independent concurrency controls over shared data. That rationale explains isolate the workloads onto separate virtual warehouses and tune each independently. It does not establish the mechanism needed here: Snowflake does not promise reuse of a persisted result after the underlying result-relevant data changes.
C:Poor partition pruning forces more data scanning even when returned row count is small. That rationale explains inspect pruning statistics and the data distribution before choosing clustering or predicate changes. It does not establish the mechanism needed here: Snowflake does not promise reuse of a persisted result after the underlying result-relevant data changes.
D:Aggregate correctness depends on matching the intended grain, not merely using SUM after an exploding join. That rationale explains repair join cardinality and then apply the correct aggregation grain. It does not establish the mechanism needed here: Snowflake does not promise reuse of a persisted result after the underlying result-relevant data changes.
Question 278
A very large fact table is filtered repeatedly by a selective date range but its partitions overlap heavily. What is the most appropriate next design or operational decision?
- Use a supported IDE integration with authenticated Snowflake connectivity and checked-in SQL files.
- Evaluate a clustering key aligned with the common date filter and monitor reclustering cost.
- Explain that Snowflake stores persisted table data in a storage layer separate from virtual warehouse compute.
- Create an appropriate schema within the intended database and grant only necessary object privileges.
Explanation
Correct Answer
B:Clustering helps related values reside in fewer micro-partitions, supporting stronger pruning. Success should be confirmed by this scenario-specific evidence: similar filtered queries scan fewer partitions after a beneficial reclustering.
Incorrect Answers
A:An editor integration keeps source files in the development workflow while executing SQL against the chosen Snowflake account. In contrast, clustering helps related values reside in fewer micro-partitions, supporting stronger pruning.
C:Independent storage and compute let a query warehouse change size while the same durable table data remains centrally managed. In contrast, clustering helps related values reside in fewer micro-partitions, supporting stronger pruning.
D:Schemas are database-level namespaces for organizing database objects and access. In contrast, clustering helps related values reside in fewer micro-partitions, supporting stronger pruning.
Question 279
A team is debugging SQL run by different clients that appear to use unexpected databases or schemas. The team proposes this action: Inspect the session's active database, schema, role, and warehouse context before changing SQL. Why should that action address the stated requirement?
- Schemas are database-level namespaces for organizing database objects and access.
- Multi-cluster warehouse availability is edition-sensitive; higher compute demand alone does not add gated features to Standard Edition.
- Unqualified object names and available privileges are interpreted using active session context.
- Auto-resume starts an eligible suspended warehouse when a session submits work requiring that warehouse.
Explanation
Correct Answer
C:Unqualified object names and available privileges are interpreted using active session context. This explains the proposed action: inspect the session's active database, schema, role, and warehouse context before changing SQL.
Incorrect Answers
A:Schemas are database-level namespaces for organizing database objects and access. That rationale explains create an appropriate schema within the intended database and grant only necessary object privileges. It does not establish the mechanism needed here: unqualified object names and available privileges are interpreted using active session context.
B:Multi-cluster warehouse availability is edition-sensitive; higher compute demand alone does not add gated features to Standard Edition. That rationale explains check edition entitlements and choose an edition that includes multi-cluster warehouses. It does not establish the mechanism needed here: unqualified object names and available privileges are interpreted using active session context.
D:Auto-resume starts an eligible suspended warehouse when a session submits work requiring that warehouse. That rationale explains enable AUTO_RESUME when the workflow should start compute automatically on demand. It does not establish the mechanism needed here: unqualified object names and available privileges are interpreted using active session context.
Question 280
A Snowflake stored procedure must contact an approved external HTTPS service via a managed integration. Before implementing the planned approach, which prerequisite should be checked most directly?
- Check stage type, directory-table configuration and any refresh requirements.
- Check destination permissions, file format, partition strategy, and access to the source data.
- Agree on the tolerated failure threshold and retain load status and rejection evidence.
- Check the integration type, endpoint authorization, secrets and execution-role grants.
Explanation
Correct Answer
D:Check the integration type, endpoint authorization, secrets and execution-role grants. The condition matters because integrations govern supported network and external-service access from Snowflake-managed code.
Incorrect Answers
A:Check stage type, directory-table configuration and any refresh requirements. That preparatory check belongs to a document workflow stores PDFs in object storage and needs staged file metadata for cataloging and processing. Here, the decisive check is check the integration type, endpoint authorization, secrets and execution-role grants.
B:Check destination permissions, file format, partition strategy, and access to the source data. That preparatory check belongs to an external partner needs a scheduled extract of approved table rows written into cloud storage. Here, the decisive check is check the integration type, endpoint authorization, secrets and execution-role grants.
C:Agree on the tolerated failure threshold and retain load status and rejection evidence. That preparatory check belongs to a batch load should tolerate a small number of malformed records while retaining the rows that parse correctly. Here, the decisive check is check the integration type, endpoint authorization, secrets and execution-role grants.
SnowPro Core COF-C03 Practice Tests