We break OpenStack. We break all of OpenStack. All the time. Why? Because we deliver OpenStack to the Enterprise so that the Enterprise can run large scale applications with tight SLAs, and the only way to achieve that level of reliability is to break everything all the time -- understand the impacts of the breakage.
In this talk, we will share what we've learned for the benefit of the community, and of OpenStack itself. We'll talk about how we:
- Build the tools, benchmarks, processes, architecture and extensions to enable OpenStack to be used for large enterprise mission critical use cases
- Indoctrinate the product engineering team with this information so that it can be translated into the required changes for Mirantis Openstack products
- Publish the verified results widely to build confidence in OpenStack.
Our goal is to provide everything necessary to run performance applications at large scales with no user visible failures. We will present our preliminary performance and reliability results for enterprise applications across compute, networking -- that's Neutron, NOT nova network -- and storage at scales of larger than 100 hypervisors.
We will include a section on the tools and operational processes necessary for the resiliency and performance achieved, as well as present our roadmap for greater scale, more application types and higher resiliency and performance.