Book Image

OpenStack Sahara Essentials

By : Omar Khedher
Book Image

OpenStack Sahara Essentials

By: Omar Khedher

Overview of this book

The Sahara project is a module that aims to simplify the building of data processing capabilities on OpenStack. The goal of this book is to provide a focused, fast paced guide to installing, configuring, and getting started with integrating Hadoop with OpenStack, using Sahara. The book should explain to users how to deploy their data-intensive Hadoop and Spark clusters on top of OpenStack. It will also cover how to use the Sahara REST API, how to develop applications for Elastic Data Processing on Openstack, and setting up hadoop or spark clusters on Openstack.
Table of Contents (14 chapters)

Defining the network


Apparently, one of the most pertinent questions that pops up when designing an OpenStack environment is the network setup and configuration. Networking in OpenStack might appear complex when we consider using Neutron instead of the Nova-network service. Without going into too much depth by distinguishing the difference between both network services in OpenStack, it might be crucial to identify how Sahara can handle the network operation in a Hadoop cluster using either Nova-network or neutron. By default, Sahara uses nova-network. However, we have configured in our RDO installation to use Neutron. This can be verified in the /etc/sahara/sahara-api.conf configuration file in the cloud controller:

…
use_neutron=True
…

Most importantly, Sahara requires SSH access to the instances of the Hadoop cluster once successfully spawned. To do so, Sahara will use either their fixed or floating IP.

By default, Sahara uses floating IP. This can be verified in the Sahara configuration...