Learn Snowflake. Build with confidence.

Snowflake is a cloud-based data warehouse platform designed to store and analyze large volumes of data without requiring you to manage physical hardware.

The separation of storage and compute allows you to scale them independently and pay for compute based on actual usage.

A warehouse is Snowflake's compute engine. It provides CPU and memory required to execute queries and can automatically suspend when idle.

CREATE WAREHOUSE my_wh
  WAREHOUSE_SIZE = 'XSMALL'
  AUTO_SUSPEND = 60
  AUTO_RESUME = TRUE;

A database is the top-level container used to organize schemas, tables, views and other database objects.

CREATE DATABASE sales_db;

USE DATABASE sales_db;

A schema sits inside a database and groups related tables, views and other objects together. Think of it like a folder inside a database.

CREATE SCHEMA sales_db.raw_data;

Tables contain the actual rows and columns of your data, similar to tables in traditional relational databases.

CREATE TABLE customers (
  id INT,
  name STRING,
  email STRING,
  signup_date DATE
);

A stage is a storage location where files are kept before or after loading them into Snowflake. Stages can be internal or external.

CREATE STAGE ext_stage
  URL = 's3://my-bucket/data/'
  CREDENTIALS = (
    AWS_KEY_ID='***'
    AWS_SECRET_KEY='***'
  );

Snowflake supports standard ANSI SQL. Common commands such as SELECT, JOIN, GROUP BY, WHERE and ORDER BY work as expected.

SELECT
    customer_id,
    SUM(amount) AS total_sales
FROM sales
GROUP BY customer_id
ORDER BY total_sales DESC;

A file format tells Snowflake how incoming files should be interpreted. It can define delimiters, headers and other file properties.

CREATE FILE FORMAT csv_fmt
  TYPE = 'CSV'
  FIELD_DELIMITER = ','
  SKIP_HEADER = 1;

Snowflake provides different approaches for loading data. Two important approaches are Bulk Loading and Snowpipe continuous loading.

COPY INTO loads a batch of files from a stage into a Snowflake table.

COPY INTO customers
FROM @my_stage
FILE_FORMAT = (FORMAT_NAME = csv_fmt)
ON_ERROR = 'CONTINUE';

Snowpipe automatically loads new files into Snowflake when they arrive in cloud storage, making it useful for continuous data ingestion.

CREATE PIPE my_pipe
  AUTO_INGEST = TRUE
AS
COPY INTO customers
FROM @my_stage
FILE_FORMAT = (FORMAT_NAME = csv_fmt);

A Stream tracks row-level changes such as inserts, updates and deletes made to a table, allowing incremental processing.

CREATE STREAM orders_stream
ON TABLE orders;

A Task automatically executes SQL statements on a schedule. Tasks are often combined with Streams to automate transformations.

CREATE TASK my_task
  WAREHOUSE = my_wh
  SCHEDULE = '5 MINUTE'
AS
INSERT INTO summary
SELECT * FROM orders_stream;

Time Travel allows you to query, restore or recover historical versions of data within Snowflake's configured retention period.

SELECT *
FROM orders
AT (OFFSET => -3600);

UNDROP TABLE orders;

Now combine everything you've learned into a real data engineering pipeline.