Snowflake Data Cloud Cheat Sheet

Updated 2026-05-15

Next Topic: Snowflake Streams Tasks and Dynamic Tables Cheat Sheet

🧠Study flashcards on this topic122 cards · spaced repetition→

Snowflake Data Cloud is a cloud-native data platform that separates storage and compute, enabling elastic scaling for analytics, data engineering, and AI workloads across AWS, Azure, and Google Cloud. Unlike traditional databases, Snowflake's micro-partition architecture automatically organizes data for optimal pruning, while features like Time Travel, Zero-Copy Cloning, and Secure Data Sharing redefine collaboration and recovery. Its serverless model — from Snowpipe ingestion to Cortex AI functions — eliminates infrastructure management, making Snowflake a unified platform for warehousing, data lakes, and streaming.

What This Cheat Sheet Covers

This topic spans 25 focused tables and 132 indexed concepts, 122 flashcards. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.

Table 1: Virtual Warehouse ConfigurationTable 2: Micro-Partitions and ClusteringTable 3: Time Travel and Fail-SafeTable 4: Zero-Copy CloningTable 5: Continuous Data IngestionTable 6: Streams and TasksTable 7: Snowpark Python APITable 8: Cortex AI FunctionsTable 9: Secure Data SharingTable 10: Data Governance PoliciesTable 11: Query Optimization FeaturesTable 12: Advanced Table TypesTable 13: Data Loading and UnloadingTable 14: File Formats and StagesTable 15: Security and Network ControlsTable 16: Resource Monitoring and Cost ControlTable 17: Data Replication and AvailabilityTable 18: Data Clean Rooms and CollaborationTable 19: Metadata and Information SchemaTable 20: Geospatial FunctionsTable 21: Advanced Query ConstructsTable 22: User-Defined ObjectsTable 23: Sequences and IdentityTable 24: Data Sampling and ProfilingTable 25: Storage Lifecycle and Archival

Quick IndexSubscribe to unlock

A jump-to index of every table row in this cheat sheet.

Mind MapSubscribe to unlock

An interactive map of every table and concept in this topic.

Table 1: Virtual Warehouse Configuration

A virtual warehouse is the compute that actually runs your queries, and almost every cost decision in Snowflake starts here. These parameters control how much horsepower a warehouse has, how it scales out under concurrency, and — crucially — how quickly it suspends itself when idle, which is the single biggest lever you have on the credit bill.

Parameter	Example	Description
Warehouse Size	`CREATE WAREHOUSE wh_medium` `WAREHOUSE_SIZE = 'MEDIUM';`	• Defines compute capacity: X-Small to 6X-Large — each size doubles credits/hour and parallelism • choose based on query complexity and concurrency
Multi-Cluster Mode	`ALTER WAREHOUSE wh SET` `MIN_CLUSTER_COUNT = 2` `MAX_CLUSTER_COUNT = 8;`	• Scales horizontally by adding clusters for concurrent queries • MIN/MAX define scaling boundaries • AUTO_SUSPEND handles idle clusters
Scaling Policy	`ALTER WAREHOUSE wh SET` `SCALING_POLICY = 'ECONOMY';`	• STANDARD starts clusters immediately • ECONOMY favors queuing to minimize cost — choose based on latency vs. budget priority
Auto-Suspend	`CREATE WAREHOUSE wh` `AUTO_SUSPEND = 60;`	• Suspends warehouse after N seconds of inactivity — critical for cost control • 60s is aggressive, 600s balances cache retention
Auto-Resume	`CREATE WAREHOUSE wh` `AUTO_RESUME = TRUE;`	• Automatically starts warehouse when query submitted — enabled by default • disable for manual control scenarios
Statement Timeout	`ALTER WAREHOUSE wh SET` `STATEMENT_TIMEOUT_IN_SECONDS = 3600;`	• Kills queries exceeding time limit — prevents runaway resource consumption • set per warehouse or session
Query Queue Timeout	`ALTER WAREHOUSE wh SET` `STATEMENT_QUEUED_TIMEOUT_IN_SECONDS = 1800;`	Cancels queries waiting in queue beyond threshold — useful when multi-cluster MAX reached and new queries pile up.

Table 2: Micro-Partitions and Clustering

Snowflake silently splits every table into immutable micro-partitions and records min/max stats per column, so most queries skip the bulk of your data without you doing anything. Clustering is the optional next step — when a big table is filtered hard on one or two columns, co-locating those values lets pruning go even further, at the cost of background credits.

Concept	Example	Description
Micro-Partition	Automatic on `INSERT`	• Contiguous 50-500 MB units (uncompressed) storing columnar data • Snowflake tracks min/max per column for pruning — no user action required

The rest of this section is part of the full cheat sheet

Table 3: Time Travel and Fail-Safe

Because micro-partitions are immutable, Snowflake can hand you the table exactly as it looked seconds, hours, or days ago — query the past, undrop a table someone deleted, or clone from a point before a bad load. Time Travel is the window you control (0–90 days); Fail-Safe is the seven-day backstop beyond it that only Snowflake Support can reach into, meant for disasters rather than everyday mistakes.

Feature	Example	Description
Time Travel Query (AT)	`SELECT * FROM t1 AT (OFFSET => -3600);`	• Accesses data as it existed N seconds ago — uses AT with TIMESTAMP or OFFSET • retention period configurable 0-90 days

The rest of this section is part of the full cheat sheet

Table 4: Zero-Copy Cloning

Cloning a table, schema, or whole database is instant and free at creation — the clone just shares the same micro-partitions through metadata pointers, and storage only grows as the two copies diverge. This is what makes spinning up a full-scale dev or test environment from production a one-line operation rather than an overnight copy job.

Operation	Example	Description
Clone Table	`CREATE TABLE t2 CLONE t1;`	• Instantly duplicates table without copying data — metadata pointers share micro-partitions • changes diverge after creation

The rest of this section is part of the full cheat sheet

Table 1: Virtual Warehouse Configuration

Parameter	Example	Description
Warehouse Size	`CREATE WAREHOUSE wh_medium` `WAREHOUSE_SIZE = 'MEDIUM';`	• Defines compute capacity: X-Small to 6X-Large — each size doubles credits/hour and parallelism • choose based on query complexity and concurrency
Multi-Cluster Mode	`ALTER WAREHOUSE wh SET` `MIN_CLUSTER_COUNT = 2` `MAX_CLUSTER_COUNT = 8;`	• Scales horizontally by adding clusters for concurrent queries • MIN/MAX define scaling boundaries • AUTO_SUSPEND handles idle clusters
Scaling Policy	`ALTER WAREHOUSE wh SET` `SCALING_POLICY = 'ECONOMY';`	• STANDARD starts clusters immediately • ECONOMY favors queuing to minimize cost — choose based on latency vs. budget priority
Auto-Suspend	`CREATE WAREHOUSE wh` `AUTO_SUSPEND = 60;`	• Suspends warehouse after N seconds of inactivity — critical for cost control • 60s is aggressive, 600s balances cache retention
Auto-Resume	`CREATE WAREHOUSE wh` `AUTO_RESUME = TRUE;`	• Automatically starts warehouse when query submitted — enabled by default • disable for manual control scenarios
Statement Timeout	`ALTER WAREHOUSE wh SET` `STATEMENT_TIMEOUT_IN_SECONDS = 3600;`	• Kills queries exceeding time limit — prevents runaway resource consumption • set per warehouse or session
Query Queue Timeout	`ALTER WAREHOUSE wh SET` `STATEMENT_QUEUED_TIMEOUT_IN_SECONDS = 1800;`	Cancels queries waiting in queue beyond threshold — useful when multi-cluster MAX reached and new queries pile up.

Table 2: Micro-Partitions and Clustering

Concept	Example	Description
Micro-Partition	Automatic on `INSERT`	• Contiguous 50-500 MB units (uncompressed) storing columnar data • Snowflake tracks min/max per column for pruning — no user action required

Table 3: Time Travel and Fail-Safe

Feature	Example	Description
Time Travel Query (AT)	`SELECT * FROM t1 AT (OFFSET => -3600);`	• Accesses data as it existed N seconds ago — uses AT with TIMESTAMP or OFFSET • retention period configurable 0-90 days

Table 4: Zero-Copy Cloning

Operation	Example	Description
Clone Table	`CREATE TABLE t2 CLONE t1;`	• Instantly duplicates table without copying data — metadata pointers share micro-partitions • changes diverge after creation