Explore the Apache Spark file format ecosystem in this 29-minute video from Databricks. Dive into the critical role of storage and IO in Spark job performance and optimization. Learn about key file formats like Parquet, ORC, and Avro, their origins, core data structures, and optimization techniques. Discover how to tune SparkConf and SQLConf settings for these formats. Examine real-world industry examples showcasing file format impact on job performance and stability. Gain insights into emerging technologies like Apache Arrow. By the end, understand core concepts of prevalent file formats, relevant settings, and how to select the right format for your Spark jobs, especially for IO-intensive AI/ML workflows.
Overview
Syllabus
Intro
Session Goals
File Formats
Row-wise Storage
Columnar (Column-wise) Storage
Hybrid Storage
Example Data
About: CSV
About: JSON
About: Avro
Inspecting: Avro
About: ORC
Structure: ORC
Inspecting: ORC
Config: ORC
Structure: Parquet
Inspecting: Parquet (1)
Inspecting: Parquet (2)
Config: Parquet
Case Study: Veraset
Looking Forward: Apache Arrow
Final Thoughts
Taught by
Databricks
Related Courses
-
Distributed Computing with Spark SQL
-
The Parquet Format and Performance Optimization Opportunities
-
Optimizing Spark SQL Jobs with Parallel and Asynchronous IO
-
Recent Parquet Improvements in Apache Spark - Vectorized Complex Types and Column Index Support
-
Degrading Performance - Understanding and Solving Small Files Syndrome
-
Fixing Small Files Performance Issues in Apache Spark Using DataFlint
