Whilst HA of the OpenStack control plane is mostly a solved problem, HA for guest VMs and compute nodes is still a work in progress, with several partially overlapping implementations available based on masakari, Pacemaker, Mistral, and others.
This talk from core members of the OpenStack HA team presents the latest news on a project to converge on a single upstream solution which is modular, flexible, and provides a smooth migration path from already deployed solutions. This project aims to address the User Story defined and prioritised by the Product Working Group in order to address the common requirement for operators who need to run legacy "pet" workloads within their OpenStack clouds.
We will also explain how developers and operators can get involved in this project in order to provide guest HA in their own clouds.
Attendees can expect to learn:
- What High Availability of the compute plane (compute nodes, hypervisors, and guest VMs) entails
- Why High Availability of the compute plane is important
- A summary of the solutions currently available upstream
- The status and future plans of the project to converge these solutions
- How they can get involved with the project