A recently proposed spec to Cinder suggests using a database to avoid raciness when multiple schedulers are used in the Cinder environment. The described raciness is the result of each scheduler having an independent view of the available resources in the environment. This problem can be solved by having a single authoritative service for tracking resource inventory and usage.
Placement does this.
This session will bring Placement and Cinder community members together to determine how Cinder can use Placement. We will:
- Help Placement developers understand how scheduling currently works in Cinder.
- Help Cinder developers understand what Placement can do.
- Develop short and medium term plans to get started with making it real.
This session will also be great for other projects interested in learning how Placement can be leveraged for some of their needs, and hopefully provide an example of how to introduce Placement into their projects.