Book Image

Data Modeling with Snowflake

By : Serge Gershkovich
5 (2)
Book Image

Data Modeling with Snowflake

5 (2)
By: Serge Gershkovich

Overview of this book

The Snowflake Data Cloud is one of the fastest-growing platforms for data warehousing and application workloads. Snowflake's scalable, cloud-native architecture and expansive set of features and objects enables you to deliver data solutions quicker than ever before. Yet, we must ensure that these solutions are developed using recommended design patterns and accompanied by documentation that’s easily accessible to everyone in the organization. This book will help you get familiar with simple and practical data modeling frameworks that accelerate agile design and evolve with the project from concept to code. These universal principles have helped guide database design for decades, and this book pairs them with unique Snowflake-native objects and examples like never before – giving you a two-for-one crash course in theory as well as direct application. By the end of this Snowflake book, you’ll have learned how to leverage Snowflake’s innovative features, such as time travel, zero-copy cloning, and change-data-capture, to create cost-effective, efficient designs through time-tested modeling principles that are easily digestible when coupled with real-world examples.
Table of Contents (24 chapters)
1
Part 1: Core Concepts in Data Modeling and Snowflake Architecture
8
Part 2: Applied Modeling from Idea to Deployment
14
Part 3: Solving Real-World Problems with Transformational Modeling

Streams

Streams are logical objects that capture data changes in underlying sources, including the previously mentioned objects (physical tables, views, and external and directory tables). Whenever a DML operation occurs in the source object, a stream tracks the changes (inserts, deletions, and the before/after images of updates). Streams achieve this through an offset storage technique—logically taking an initial snapshot of data and then tracking changes through metadata columns. Although a stream can be queried like a table, it is not a separate object and does not contain table data.

When a stream is created, metadata columns are tacked onto the source object and begin tracking changes. The following table describes the metadata fields and their contents:

Figure 4.4 – Stream metadata columns

Figure 4.4 – Stream metadata columns

The following command creates a stream on a table:

CREATE STREAM <stream_name> ON TABLE <table_name>

For every subsequent DML...