[Underscore] 2018-05: Async with Akka, Spark Pitfalls

We'd like to invite you to our May meetup!

Keep sending us your talk and improvement suggestions, we're always on the lookout for someone looking to share their experience! (Call for talks here: https://docs.google.com/document/d/1YbnPsqfyO-CDwvlwsq5cGLFUTXnkoRTwvBE6bq_MBvY)

Session #1: Writing Asynchronous Programs with Scala and Akka - Yardena Meymann
Session #2: Spark Pitfalls - Lior Regev

Sessions will be recorded.
We will be graciously hosted by BigPanda, who will be providing food and drinks.

== Writing Asynchronous Programs with Scala and Akka ==

This talk will present the multiple options Scala and Akka provide to write asynchronous code, from Futures and Promises, to Actors and Streams and how to combine them. We will look at some practices we developed at feature.fm regarding when to use each of the tools in our quest to create a 100% non-blocking backend, present some problems we encountered and the solutions/tricks we came up with to overcome them. We will cover examples of using libraries like reactive-mongo, Akka HTTP and reactive-kafka.

Yardena is a senior software developer, architect and trainer with over 20 years experience in the industry, over decade in Java and recent 5 years almost exclusively Scala. After working on large enterprise projects at Cisco, HP and VMware, she joined Feature.fm start-up 3 years ago to build its backend infrastructure from scratch. During this period Yardena acquired a lot of experience with Akka, Kafka and Spark. Yardena is also a programming language enthusiast and one of the organizers of Sayeret Lambda meet-up.

== Spark Pitfalls ==

Apache Spark is an increasingly popular framework for processing big data in a fast, resilient manner.
One of the properties of Spark is the attempt to run mostly in-memory as opposed to saving every step to a filesystem.
This, and other properties of Spark create a pool of interesting pitfalls that can cause Spark applications to run poorly or even crash.
In this talk, I intend to share my experience with various pitfalls and workarounds as well as some optimization tips for Spark applications.

Lior Regev is a lead software developer at Endor. He has been developing Spark applications for the past couple of years, on clusters ranging from 2 to 500 nodes and optimizing them to run smoothly, quickly and deterministically.