Book Image

Fast Data Processing Systems with SMACK Stack

By : Raúl Estrada
Book Image

Fast Data Processing Systems with SMACK Stack

By: Raúl Estrada

Overview of this book

SMACK is an open source full stack for big data architecture. It is a combination of Spark, Mesos, Akka, Cassandra, and Kafka. This stack is the newest technique developers have begun to use to tackle critical real-time analytics for big data. This highly practical guide will teach you how to integrate these technologies to create a highly efficient data analysis system for fast data processing. We’ll start off with an introduction to SMACK and show you when to use it. First you’ll get to grips with functional thinking and problem solving using Scala. Next you’ll come to understand the Akka architecture. Then you’ll get to know how to improve the data structure architecture and optimize resources using Apache Spark. Moving forward, you’ll learn how to perform linear scalability in databases with Apache Cassandra. You’ll grasp the high throughput distributed messaging systems using Apache Kafka. We’ll show you how to build a cheap but effective cluster infrastructure with Apache Mesos. Finally, you will deep dive into the different aspect of SMACK using a few case studies. By the end of the book, you will be able to integrate all the components of the SMACK stack and use them together to achieve highly effective and fast data processing.
Table of Contents (15 chapters)
Fast Data Processing Systems with SMACK Stack
Credits
About the Author
About the Reviewers
www.PacktPub.com
Preface

Spark Mesos run modes


Spark can run over Mesos in two modes: coarse-grained (default) and fine-grained (deprecated).

Coarse-grained

In coarse-grained mode, each Spark executor runs as a single Mesos task. Spark executors are sized according to the following configuration variables:

  • Executor memory: spark.executor.memory
  • Executor cores: spark.executor.cores
  • Number of executors: spark.cores.max/spark.executor.cores

Executors are brought up when the application starts, until spark.cores.max is reached. If spark.cores.max is not set, the Spark application will reserve all resources offered to it by Mesos, so it is highly recommended to set this variable in any sort of multi-tenant cluster, including those running multiple concurrent Spark applications.

The scheduler will start the executors round-robin on the offers Mesos gives it, but there are no spread guarantees, as Mesos does not provide such guarantees on the offer stream.

The benefit of the coarse-grained mode is a much lower startup overhead...