Hands-On With Spark: Creating a Fast Data Pipeline

A Tale of Two Streaming APIs

Fast Data architectures have emerged as the answer for enterprises that need to process and analyze continuous streams of data. Apache Spark has matured into a very popular framework for data analytics that — when combined with other technologies found in Lightbend Fast Data Platform like Akka Streams, Kafka, and Mesos — helps businesses accelerate decision making and become reactive to the particular characteristics of their market.

Spark combines various libraries like SQL-based analytics, Fast Data flow processing, graph analytics, and a rich library of built-in Machine Learning algorithms to address a wide range of requirements for large-scale data analytics. But how can you know which part to use for the right job?

In this talk by Gerard Maas, O'Reilly author and Senior Software Engineer at Lightbend, we focus on choosing the right Fast Data stream processing features of Apache Spark, taking a practical, code-driven look at the two APIs available for this: the mature Spark Streaming and its younger sibling, Structured Streaming. Specifically, we will review:

• The capabilities of Spark's APIs for streaming and their key differences
• Advice on making the right choice for an application and how to architect and develop streaming pipelines that use one or both APIs to fulfill its requirements
• A code-based demo on getting started with both the Spark Streaming and Structured Streaming APIs

Watch the Full Video + Demo (~60 Min.)

If you would like to continue learning, here are some curated resources from our webinar presenter:

• Demo Materials: the Spark Notebooks used in this presentation
• From Reactive Summit 2018: Taking The Pain Out Of Deploying Streaming Applications
• Free O'Reilly eBook: Designing Fast Data Application Architectures