Snowflake

Snowflake Glossary: 80 Essential Exam Terms Organized by Category

2026-03-26
更新: 2026-03-27
NicheeLab Editorial Team

SnowPro exam questions are packed with specialized terminology: the three-layer architecture, Virtual Warehouse, RBAC, Time Travel, Data Sharing, and more. Accurate understanding of terminology is the foundation of passing. This article curates 80 essential exam terms and organizes them into 5 categories — Architecture, SQL, Security, Data Sharing, and Performance — explained in scannable table format.

Architecture Terms (16 entries)

Terms covering Snowflake's three-layer architecture and platform composition. This is the highest-frequency topic on the SnowPro Core exam, weighted at 25-30% in Domain 1.

No.TermDescription
1Snowflake AI Data CloudUnifies data warehousing, data lakes, data sharing, and ML/AI workloads on a single platform. Runs on AWS, Azure, and GCP.
2Three-Layer ArchitectureStorage Layer, Compute Layer, and Cloud Services Layer are fully decoupled and scale independently. Snowflake's defining characteristic.
3Storage LayerHandles persistent storage. Data is compressed, encrypted, and stored on the cloud provider's object storage (S3 / Azure Blob / GCS) as micro-partitions.
4Compute LayerHandles query execution. Provides independent compute clusters as Virtual Warehouses, fully decoupled from storage.
5Cloud Services LayerHandles metadata management, authentication, authorization, query optimization, and transaction management. Result Cache and Metadata Cache also live here.
6Micro-partitionThe unit of data storage. Columnar 50-500 MB compressed data blocks. Automatically managed by Snowflake — no user definition required. Their metadata drives pruning.
7Virtual WarehouseThe unit of compute resources, sized from X-Small to 6X-Large. Each size step roughly doubles resources. Supports auto-suspend/resume.
8Multi-cluster WarehouseEnterprise edition and above. Two modes: Auto-scaling (clusters scale with load) and Maximized (always at maximum clusters). Handles concurrent query workloads.
9Snowflake EditionsFour tiers: Standard, Enterprise, Business Critical, and VPS. Higher tiers add security features and have higher credit unit pricing.
10CreditThe unit of compute consumption. Calculated as Warehouse runtime × size. Also consumed by Serverless features and excessive Cloud Services usage.
11SnowparkFramework for executing data processing in Python, Java, and Scala. Used to build DataFrame API code, UDFs, and Stored Procedures.
12Snowflake MarketplaceA marketplace for discovering and acquiring third-party datasets and services. Access external data directly with no data copy.
13Snowflake CortexUmbrella name for Snowflake's AI/ML capabilities. Includes LLM functions (COMPLETE / TRANSLATE / SUMMARIZE), ML Functions (FORECAST / ANOMALY_DETECTION), and Search features.
14Native App FrameworkFramework for building and distributing applications on Snowflake. Composed of an Application Package (manifest.yml, setup.sql, Stage).
15SnowgridSnowflake's global data network. The foundation for cross-region and cross-cloud replication, failover, and Data Sharing.
16SnowsightThe web-based UI. Provides SQL query execution, visualization, dashboards, and account management. The successor to the legacy Classic Console.

SQL Terms (16 entries)

Snowflake-specific SQL extensions and data manipulation terms. Frequently tested on the Data Transformations and Data Analyst exams.

No.TermDescription
17QUALIFY clauseA Snowflake-specific SQL extension that filters on window function results. Lets you filter ROW_NUMBER() output directly without a subquery.
18FLATTEN functionA table function that expands VARIANT, ARRAY, and OBJECT values into rows. Used in the form TABLE(FLATTEN(input => data)), often combined with a LATERAL clause.
19VARIANT typeA general-purpose type for semi-structured data (JSON, XML, Avro, Parquet, etc.). Up to 16 MB. Accessed via colon notation (data:key:subkey).
20COPY INTOSQL for bulk loading and unloading data between stages and tables. Flexibly configured via FILE FORMAT, ON_ERROR, and VALIDATION_MODE options.
21Stored ProcedureWritten in JavaScript, Python, Java, or Scala. Implements DDL/DML execution, branching, loops, and exception handling. Invoked via the CALL statement.
22UDF / UDTFUser-defined functions. Scalar (UDF) and table (UDTF) variants. Can be written in SQL, JavaScript, Python, or Java.
23StreamsA CDC object that captures table changes (INSERT/UPDATE/DELETE). Three types: Standard, Append-only, and Insert-only.
24TasksAn object that runs SQL statements or stored procedures on a schedule, defined by CRON or minute interval. Dependencies are defined via task trees.
25SnowpipeA serverless service that auto-loads data when files land in a stage. Triggered by event notifications (SQS / Event Grid / Pub/Sub).
26Dynamic TableDeclaratively defined with target_lag controlling refresh frequency. Auto-incrementally refreshes as source data changes. An alternative to Streams + Tasks.
27External TableA read-only table that queries files on an external stage directly. Lets you analyze data on S3 / Azure / GCS without loading it first.
28Materialized ViewA view whose query results are physically stored and automatically incrementally refreshed. Enterprise edition and above. Used to accelerate complex aggregations.
29MERGE INTOSQL for upsert (UPDATE + INSERT) operations. Matches source and target on join conditions, branching via WHEN MATCHED / NOT MATCHED clauses.
30PIVOT / UNPIVOTPIVOT turns rows into columns; UNPIVOT turns columns into rows. Used for cross-tabulation and wide/long format conversions.
31Window functionAggregates, ranks, and computes moving values per partition via the OVER clause. ROW_NUMBER, RANK, LAG, LEAD, etc. Often combined with QUALIFY.
32SequenceAn object that generates unique numbers. Used to generate surrogate keys. Get the next value with my_seq.NEXTVAL.

Security Terms (16 entries)

Terms covering access control, authentication, and encryption. Tested in Domain 2 (Account Access & Security / 20-25%).

No.TermDescription
33RBACRole-Based Access Control. Privileges are granted to roles, and roles are assigned to users. Snowflake's primary access control model.
34DACDiscretionary Access Control. Object owners grant privileges to other roles. The creating role automatically becomes the OWNER.
35ACCOUNTADMINThe top-level system-defined role. Combines SYSADMIN and SECURITYADMIN privileges. Not recommended for daily use. MFA strongly recommended.
36SYSADMINRole for creating and managing databases and Warehouses. Recommended for day-to-day object management.
37SECURITYADMINRole responsible for security management of roles, privileges, and users. Owns GRANT / REVOKE administration.
38USERADMINRole specialized for creating and managing users and roles.
39Network PolicyIP-address-based access control. Restricts source IPs via ALLOWED_IP_LIST and BLOCKED_IP_LIST. Applied at the account or user level.
40MFAMulti-Factor Authentication using Duo Mobile. Strongly recommended for ACCOUNTADMIN. Can also be enforced at the account level.
41SSOSingle Sign-On based on SAML 2.0. Integrates with Okta, Azure AD, and ADFS. Federated Authentication.
42Row Access PolicyRow-level access control. Automatically filters visible rows based on the querying user's role. Enterprise edition and above.
43Dynamic Data MaskingColumn-level data masking. Controls visibility of sensitive data based on the querying role. Enterprise edition and above.
44Tri-Secret SecureBusiness Critical edition and above. Composite encryption using a Snowflake-managed key plus a customer-managed key (KMS). Customers can block access by disabling their key.
45Object TaggingTag objects to classify data. Tag-based Masking Policies enable automatic, tag-driven masking.
46Snowflake ScriptingA SQL-based procedural extension. Supports IF/ELSE, LOOP, FOR, WHILE, and TRY-CATCH inside Stored Procedures. Also supports Anonymous Blocks.
47AWS PrivateLinkBusiness Critical edition and above. Private connectivity that keeps Snowflake traffic inside the VPC instead of crossing the public internet.
48Aggregation PolicyA policy that enforces a minimum aggregation level on shared data. Blocks viewing individual rows and permits only group-level aggregates.

Test your terminology with practice questions

Check your level with questions covering every Snowflake exam

Try free questions

Data Sharing Terms (16 entries)

Terms covering data sharing, collaboration, and data protection. Tested in Domain 6 (Data Protection & Data Sharing).

No.TermDescription
49Secure Data SharingShare data with other Snowflake accounts without copying it. The provider creates a Share, then the consumer creates a database from it. Real-time sharing.
50ShareThe unit object for data sharing. Add tables, views, and UDFs to a Share, then GRANT to consumer accounts.
51Reader AccountA read-only account for organizations without their own Snowflake account. The provider creates and manages it and pays the compute costs.
52ListingThe unit of a data product on the Marketplace. Three types: Free, Personalized, and Paid. How providers distribute data broadly.
53Data Clean RoomAn environment where multiple organizations jointly analyze data without directly viewing each other's data. Enables cross-party analysis while preserving privacy.
54Data ExchangeAn invite-only private data sharing group. Suited for sharing only among specific members.
55Time TravelReference or restore data from any point in the past. Specified with AT / BEFORE clauses. Up to 1 day on Standard, up to 90 days on Enterprise+.
56Fail-safeA safety net for data recovery after Time Travel expires. Fixed at 7 days. Recoverable only by Snowflake support — users cannot access it directly.
57Zero-copy CloneInstantly creates a logical copy of a table, schema, or database. No physical copy occurs. Additional cost accrues only for changed data.
58ReplicationReplicates databases, schemas, and account objects across regions and clouds. Used for DR, distributed reads, and regulatory compliance.
59FailoverSwitches to the secondary when the primary fails. Business Critical edition and above. Combined with replication to implement BCP.
60DATA_RETENTION_TIME_IN_DAYSParameter that configures the Time Travel retention period. Default 1 day, up to 90 days on Enterprise+. Settable at account / database / schema / table level.
61Transient tableA table with no Fail-safe period and up to 1 day of Time Travel. Suited for staging and intermediate data. Reduces storage costs.
62Temporary tableA session-scoped table. Automatically dropped when the session ends. Invisible to other users. No Fail-safe.
63UNDROPCommand to restore a table, schema, or database dropped within the Time Travel period. Requires renaming if an object with the same name already exists.
64Secure ViewA view whose definition (SQL) is hidden. Prevents information leakage through query optimization when sharing data. Definition is not shown by SHOW VIEWS.

Performance Terms (16 entries)

Terms covering performance optimization and cost management. Tested in Domain 3 (Performance Concepts / 10-15%).

No.TermDescription
65Result CacheQuery result cache managed by the Cloud Services Layer. Returns results without a Warehouse if the same query runs again within 24 hours. Invalidated when the underlying data changes.
66Metadata CacheManaged by the Cloud Services Layer. Returns aggregates like COUNT(*), MIN, and MAX instantly from metadata. No Warehouse required.
67Warehouse CacheLocal SSD on the Compute Layer. Retains micro-partitions accessed by recent queries. Cleared when the Warehouse is suspended.
68Clustering KeyPhysical sort key for table data. Set via ALTER TABLE ... CLUSTER BY. Setting it on filter columns improves pruning efficiency. Aimed at large tables.
69Partition PruningOptimization that skips micro-partitions not matching the query predicates. Decided using min/max metadata. Applied automatically.
70Query ProfileVisualizes the query's execution plan and statistics. Shows processing time, scan volume, spilling, and pruning effectiveness. Essential for identifying bottlenecks.
71SpillingWhen memory is insufficient, data spills to local disk or remote storage. Resolved by sizing up the Warehouse.
72Resource MonitorMonitors and limits credit consumption. Sets monthly/daily caps and automatically notifies, suspends, or terminates when thresholds are reached.
73Auto-suspend / Auto-resumeAutomatically suspends when idle and resumes on a new query. Default is 10 minutes (600 seconds). A baseline cost-optimization setting.
74Search Optimization ServiceA service that accelerates point-lookup queries (equality, LIKE, IN). Automatically builds search indexes. Enterprise edition and above. Incurs additional cost.
75StageA temporary holding location for data files. Two types: internal (Snowflake-managed) and external (S3 / Azure / GCS). The source/destination for COPY INTO.
76File FormatConfigures file formatting for COPY INTO. Supports CSV, JSON, Parquet, Avro, ORC, and XML. Defines delimiters, headers, and compression.
77Storage IntegrationCentralizes credentials for external cloud storage. Consolidates IAM role and service principal configuration.
78Snowpark Container ServicesRuns Docker containers inside Snowflake. Used for GPU-accelerated deep learning and custom application hosting.
79Snowflake NotebooksJupyter-like notebooks inside Snowflake. Mix Python and SQL. Interactively run Snowpark ML and Cortex.
80Iceberg TableOpen tables in the Apache Iceberg format. The same data is also accessible from engines outside Snowflake (Spark, etc.).

Tips for Studying Terminology

1. Learn terms in contrasting pairs

Snowflake terms are frequently tested as contrasting pairs. Learning the following pairs together makes them stick and prepares you for comparison questions.

Contrasting pairKey difference
Result Cache vs Warehouse CacheCloud Services Layer vs Compute Layer. Whether it survives a suspend or not.
Time Travel vs Fail-safeUser-accessible vs accessible only via Snowflake support.
Permanent vs Transient tablesFail-safe vs no Fail-safe, with corresponding storage cost differences.
RBAC vs DACRole-mediated privilege management vs direct grants by the owner.
Scale-up vs Scale-outFaster individual queries vs higher concurrency.
Snowpipe vs COPY INTOContinuous automatic loads vs batch bulk loads.
Dynamic Table vs Streams+TasksDeclarative pipeline vs imperative pipeline.

2. Cross-reference features by edition

"Which edition is required for this feature?" is a frequent question pattern. The mappings below are must-memorize.

FeatureRequired edition
Time Travel up to 90 daysEnterprise and above
Multi-cluster WarehouseEnterprise and above
Row Access Policy / Dynamic Data MaskingEnterprise and above
Materialized ViewEnterprise and above
Search Optimization ServiceEnterprise and above
Tri-Secret SecureBusiness Critical and above
AWS PrivateLink / Azure Private LinkBusiness Critical and above
Failover / FailbackBusiness Critical and above

3. Memorize acronyms with their full spelling

Exams sometimes present acronyms in their fully spelled-out form. Memorize both the acronym and its expansion for terms like RBAC (Role-Based Access Control), DAC (Discretionary Access Control), MFA (Multi-Factor Authentication), SSO (Single Sign-On), and VPS (Virtual Private Snowflake).

Check Your Understanding

SnowPro Core - Architecture

問題 1

Which statement about Snowflake's Warehouse Cache (Local Disk Cache) is correct?

  1. Managed by the Cloud Services Layer; returns results from cache when the same query runs again within 24 hours.
  2. Held on the Compute Layer's local SSD; caches micro-partition data accessed by recent queries.
  3. A cache that returns results instantly from metadata for aggregate queries like COUNT(*) or MIN/MAX.
  4. Cache retention is 7 days and persists even when the Warehouse is suspended.

正解: B

The Warehouse Cache (Local Disk Cache) is held on the local SSD of the Compute Layer (Virtual Warehouse). It retains micro-partition data accessed by recent queries and speeds up access to that same data (B: correct). Option A describes the Result Cache. Option C describes the Metadata Cache. Option D contradicts how the Warehouse Cache actually behaves — suspending the Warehouse clears the cache. The differences between the three caches (Result / Metadata / Warehouse) in terms of management layer and retention conditions are frequently tested on the Core exam.

Frequently Asked Questions

In what order should I learn these terms?

Priority depends on which exam you're targeting. For SnowPro Core, prioritize terminology around the three-layer architecture (Storage / Compute / Cloud Services Layer), Virtual Warehouse, RBAC, and Time Travel. Domain 1 (Architecture / 25-30%) and Domain 2 (Security / 20-25%) together account for roughly 50% of the exam, so starting with these two domains is the most efficient approach. For the Advanced: Data Engineer exam, focus first on data pipelines, Streams/Tasks, Snowpipe, and Dynamic Tables.

Should I memorize technical terms in my native language or in English?

All SnowPro exams are delivered in English only, so we recommend memorizing terms in English. Proper nouns like Virtual Warehouse, Time Travel, Fail-safe, Micro-partition, Clustering Key, and Result Cache should be learned exactly as they appear. Use any translations only as a conceptual aid, and aim to instantly grasp the meaning the moment you see the English term. Mastering roughly 200-300 technical terms will let you understand most exam questions.

Can I pass the exam just by memorizing the glossary?

Memorization alone is rarely enough to pass. SnowPro exams center on conceptual understanding and applied reasoning — straight definition questions ("What is X?") are relatively rare. Most questions are comparative or judgment-based: "Which option is best in this situation?" or "What is the difference between these two features?" You need to understand not just definitions but the relationships between terms and when to use each one. Treat the glossary as a foundation, then deepen your understanding by reading the official documentation and doing hands-on practice.

Check your understanding with practice questions

Check your level with questions covering every Snowflake exam

Try free questions

Related Snowflake Certification Articles

Snowflake Certifications Overview

All 11 exams: scope, fees, and prerequisites

SnowPro Core: Complete Guide

Strategy guide for Snowflake's entry-level certification

Snowflake Exam Sample Questions

Sample questions and detailed explanations at every level

How to Study for Snowflake Certifications

Fastest path to passing and recommended study time

Check what you learned with practice questions

Practice with certification-focused question sets

無料で問題を解いてみる
Author

NicheeLab Editorial Team

NicheeLab editorial team focused on data engineering and cloud certification learning. Content is structured around practical study needs and official exam domains.


Related articles
Snowflake

Snowflake Certifications: All 11 Exams Explained (2026)

Every SnowPro certification — Associate, Core, Specialty, Ad...

Snowflake

Snowflake Exam Difficulty Ranking: All 11 Certs Compared (2026)

All 11 SnowPro exams ranked by difficulty with study-time es...

Snowflake

Snowflake Study Guide: Fastest Pass Route by Exam (2026)

How to pass SnowPro certifications efficiently — official ma...

Snowflake

SnowPro Core (COF-C03): Complete Exam Guide (2026)

Pass the SnowPro Core exam — six domains, scope, sample ques...

Snowflake

SnowPro Associate Platform (SOL-C01): Complete Guide (2026)

The entry-level SnowPro Associate exam — scope, weighting, s...

Browse all Snowflake articles (103)
© 2026 NicheeLab All rights reserved.