Deep Dive Into a Highly Available OpenStack Architecture

OpenStack High Availability

OpenStack Summit Vancouver 2015

Wednesday, May 20 • 9:00am – 9:40am

Abstract: Many organizations choose OpenStack for it’s distributed architecture, and ability to deliver Infrastructure-as-a-Service platforms for mission critical applications. In such environments, that run production workloads, the cloud control plane is ought to provide maximum possible up-time, and it is crucial to build an OpenStack cloud deployment where all OpenStack services are in a highly available configuration. Deploying a highly available control plane for OpenStack can be achieved in various configurations, each would serve certain set of demands, and introduce growing set of prerequisites.

In this talk Arthur Berezin will discuss the following:

– Overview of a highly available OpenStack architecture using Pacemaker and HAProxy.

– In-depth overview of each of OpenStack’s core services, their behaviour and readiness for to run in a highly available configuration.

– Review of new features added to the OpenStack Juno and Kilo releases to better support a complete active/active configuration.


  • HA Enabling Services
    • Pacemaker and HAProxy
  • Shared Services
    • MariaDB w/Galera, RabbitMQ w/Mirrored Queues
  • OpenStack Services
    • Keystone, Nova, Neutron, Glance, Cinder, Horizon
  • Topologies
    • Controller, Compute, Network, Storage



Share If You Care

1 Trackbacks & Pingbacks

  1. Neutron High Availability OpenStack Architecture | GREENSTACK

What do you think about this?