Snowflake SnowPro Core

Key Topics to Master:

• Cloud services layer, multi-cluster compute layer, and centralised storage layer
• Snowflake editions and which features each unlocks
• Virtual warehouse sizes, credit consumption, auto-suspend, and auto-resume
• Multi-cluster warehouses in maximized against auto-scale mode, and scaling policies
• Micro-partitions, clustering metadata, and how pruning works
• Result cache, metadata cache, and warehouse local disk cache
• Object hierarchy: organization, account, database, schema, table, stage
• Table types: permanent, transient, temporary, external, and Iceberg
• Snowsight, SnowSQL, connectors, and drivers as access paths

Key Topics to Master:

• Reading the Query Profile to identify spilling, exploding joins, and full scans
• Query history and warehouse load monitoring in Snowsight
• Scaling up against scaling out, and which one fixes which symptom
• Clustering keys, automatic clustering, and when reclustering is worth the credits
• Search optimization service and materialized views
• Querying semi-structured data with the VARIANT type, dot notation, FLATTEN, and LATERAL
• Common table expressions, window functions, and joins
• Streams and tasks for incremental transformation
• User-defined functions and stored procedures

Key Topics to Master:

• Role-based access control, role hierarchy, and privilege inheritance
• System-defined roles: ORGADMIN, ACCOUNTADMIN, SECURITYADMIN, USERADMIN, SYSADMIN, PUBLIC
• GRANT and REVOKE on objects, and future grants on schemas
• Dynamic data masking policies and external tokenization
• Row access policies and object tagging
• Network policies, multi-factor authentication, federated authentication, and SSO
• Resource monitors, credit quotas, and the actions they trigger
• ACCOUNT_USAGE and INFORMATION_SCHEMA views for auditing and cost analysis
• Access history and object dependency views

Key Topics to Master:

• Internal stages: user, table, and named, and external stages on S3, Azure Blob, or GCS
• PUT and GET for moving files to and from internal stages
• COPY INTO table for loading and COPY INTO location for unloading
• File format options including delimiters, compression, and ON_ERROR behaviour
• Snowpipe for continuous micro-batch loading and its serverless billing
• Supported file types including CSV, JSON, Avro, ORC, Parquet, and XML
• External tables over files left in cloud storage
• Storage integrations and cloud provider credentials
• Drivers and connectors, including JDBC, ODBC, Python, and SnowSQL

Key Topics to Master:

• Secure Data Sharing and how shares reference data without copying it
• Providers, consumers, and reader accounts for consumers without a Snowflake account
• Snowflake Marketplace listings and private data exchanges
• Secure views and secure UDFs in a share
• Time Travel retention periods and the differences between editions
• UNDROP for tables, schemas, and databases
• Fail-safe as a seven-day Snowflake-managed recovery window
• Zero-copy cloning of databases, schemas, and tables, and how clone storage is billed

• 1Start a Snowflake trial account and note which edition it runs on, since edition changes what Time Travel and other features allow.
• 2Read the Snowflake documentation section on key concepts and architecture, and draw the three layers from memory afterwards.
• 3Create databases, schemas, and tables and walk the object hierarchy from account downwards in Snowsight.
• 4Create permanent, transient, and temporary tables of the same data and compare what Time Travel offers on each.
• 5Download the official SnowPro Core Study Guide for COF-C03 from the Snowflake certification page and read the domain list end to end.

• 1Run the same query on an XS and on a Large warehouse, then compare runtime and credits in the query history.
• 2Set auto-suspend to 60 seconds and confirm the warehouse resumes on the next query.
• 3Configure a multi-cluster warehouse in maximized mode and then in auto-scale mode and note when each adds a cluster.
• 4Run a query twice and observe the result cache, then alter the underlying data and watch the cache miss.
• 5Create a resource monitor with a credit quota and set it to suspend the warehouse at a threshold.

• 1Read the Snowflake documentation on micro-partitions and data clustering, then run SYSTEM$CLUSTERING_INFORMATION on a large table.
• 2Compare the partitions scanned in the Query Profile for a filtered query against an unfiltered one.
• 3Define a clustering key on a table, let automatic clustering run, and re-check the clustering depth.
• 4Enable the search optimization service on one table and compare a point lookup before and after.
• 5Create a materialized view and note the maintenance cost that comes with it.

• 1Use PUT to load a local CSV to a user stage, a table stage, and a named internal stage, and list all three.
• 2Create a file format object, then run COPY INTO with ON_ERROR set to different values against a file with bad rows.
• 3Create an external stage on cloud storage with a storage integration rather than inline credentials.
• 4Set up a Snowpipe against that external stage and confirm it loads new files without a running warehouse.
• 5Create an external table over the same files and compare query behaviour against a loaded table.

• 1Load a nested JSON file into a VARIANT column, then query it with dot notation and with FLATTEN.
• 2Load Parquet and Avro files and note what changes about schema inference for each.
• 3Unload a query result with COPY INTO location and inspect the resulting files in the stage.
• 4Practise the difference between SINGLE and multiple output files, and between compressed and uncompressed unloads.
• 5Read the Snowflake documentation on the VARIANT, OBJECT, and ARRAY types and their size limits.

• 1Draw the system-defined role hierarchy from ORGADMIN down to PUBLIC and write what each role can do that the one below cannot.
• 2Create a custom role, grant it to SYSADMIN, and test that privilege inheritance works the way you drew it.
• 3Set up future grants on a schema and confirm that a table created afterwards inherits them.
• 4Apply a masking policy to a column and query it as two different roles.
• 5Apply a row access policy and confirm which rows each role sees.

• 1Query the ACCOUNT_USAGE views for warehouse credit history, login history, and query history, and note the data latency on each.
• 2Compare an ACCOUNT_USAGE view against the equivalent INFORMATION_SCHEMA view and write down the differences in retention and latency.
• 3Create a network policy and a tag, and apply the tag to a table column.
• 4Read the Snowflake documentation on MFA, federated authentication, and key pair authentication.
• 5Use ACCESS_HISTORY to trace which query touched a specific column.

• 1Set DATA_RETENTION_TIME_IN_DAYS on a table, delete rows, then recover them with AT and BEFORE clauses.
• 2Drop a table and recover it with UNDROP, then repeat for a schema and a database.
• 3Create a zero-copy clone of a database and confirm that storage is only charged as the clone diverges.
• 4Read the Snowflake documentation on Fail-safe and write down who can initiate recovery from it and over what window.
• 5Compare maximum Time Travel retention across Snowflake editions and note which edition your trial account uses.

• 1Create a share, add a database and a secure view to it, and grant it to a second account or a reader account.
• 2Write down exactly what a reader account can and cannot do, including who pays for its compute.
• 3Browse the Snowflake Marketplace, attach a free listing, and query it as a consumer.
• 4Compare a secure view against a normal view and note what the secure version hides from the consumer.
• 5Read the Snowflake documentation on sharing limitations, including which object types can be shared.

• 1Force a query to spill to local and then to remote storage and identify both in the Query Profile.
• 2Build an exploding join on purpose and recognise the row count blowup in the profile.
• 3Create a stream on a table, insert rows, and consume the stream from a task.
• 4Chain two tasks and inspect the task history for both.
• 5Write one UDF and one stored procedure and note which contexts each can be called from.

• 1Take the official SnowPro Practice Exam: Core and score yourself by domain rather than overall.
• 2Sit at least two 100-question practice sets under a strict 115-minute clock, which is 69 seconds per question.
• 3Drill multiple select questions specifically, since partial credit is not awarded and they are where scores leak.
• 4Re-study your two weakest domains, weighting time by the published percentages rather than by how uncomfortable each topic feels.
• 5Check every wrong answer against the Snowflake documentation page that covers it, since Core questions are referenceable to the docs.

• 1Review the 31 percent architecture domain one final time, since it alone is worth roughly 31 of the 100 questions.
• 2Memorise the numbers that recur: warehouse size credit multipliers, Fail-safe duration, and Time Travel maximums per edition.
• 3Register through the Snowflake certification portal and run the Pearson OnVUE system test if you are testing online.
• 4Confirm your government-issued photo ID name matches your certification profile, and remember you can reschedule an OnVUE booking free of charge any time before the appointment.
• 5Keep the following week free in case the seven-day retake wait applies.