Book Image

Business Intelligence with Databricks SQL

By : Vihag Gupta
Book Image

Business Intelligence with Databricks SQL

By: Vihag Gupta

Overview of this book

In this new era of data platform system design, data lakes and data warehouses are giving way to the lakehouse – a new type of data platform system that aims to unify all data analytics into a single platform. Databricks, with its Databricks SQL product suite, is the hottest lakehouse platform out there, harnessing the power of Apache Spark™, Delta Lake, and other innovations to enable data warehousing capabilities on the lakehouse with data lake economics. This book is a comprehensive hands-on guide that helps you explore all the advanced features, use cases, and technology components of Databricks SQL. You’ll start with the lakehouse architecture fundamentals and understand how Databricks SQL fits into it. The book then shows you how to use the platform, from exploring data, executing queries, building reports, and using dashboards through to learning the administrative aspects of the lakehouse – data security, governance, and management of the computational power of the lakehouse. You’ll also delve into the core technology enablers of Databricks SQL – Delta Lake and Photon. Finally, you’ll get hands-on with advanced SQL commands for ingesting data and maintaining the lakehouse. By the end of this book, you’ll have mastered Databricks SQL and be able to deploy and deliver fast, scalable business intelligence on the lakehouse.
Table of Contents (21 chapters)
1
Part 1: Databricks SQL on the Lakehouse
9
Part 2: Internals of Databricks SQL
13
Part 3: Databricks SQL Commands
16
Part 4: TPC-DS, Experiments, and Frequently Asked Questions

Using Serverless SQL

It would be remiss not to introduce the Serverless SQL offering in Databricks SQL. SQL Warehouses – the classic SQL Warehouses that we have discussed so far – reside in your cloud account. That is, the workers of the warehouses are VMs in your cloud account. This has two implications:

  • You incur the cost of Databricks and the cost of the VMs that Databricks uses to power the SQL Warehouses.
  • You incur some latency in cluster cold starts and upscaling events.

Databricks SQL offers Serverless SQL to circumvent this. In Serverless SQL, the SQL Warehouses are provisioned in Databricks’s account from a pool of pre-provisioned compute resources. This solves both the aforementioned implications:

  • The cost of Databricks SQL Warehouses includes the VM costs, and you pay for only one line item
  • Since Databricks uses a pool of pre-provisioned compute, no latency is incurred in cold starts, as well as when upscaling events
  • ...