Snowflake Warehouse AI Skill Guide

Overview & Engine Architecture

Snowflake separates storage (micro-partitions in cloud object storage) from compute (virtual warehouses). SQL runs on warehouses; services layer handles auth, metadata, and result cache. Roles gate objects; stages reference cloud files; COPY / Snowpipe / connectors load data. Time Travel and Fail-safe retain historical data at cost.

Stages / Snowpipe / connectors
    -> tables (micro-partitions, clustering metadata)
    -> virtual warehouse (compute credits)
    -> RESULT_SCAN / query history / RBAC

When to use / when not to

Use when