Workload segregation

The latest features of the Nova service provide different strategies to unlock a fully scalable OpenStack compute infrastructure by means of regions, Availability Zones, host aggregates, and cell concepts. To have a full picture of a large deployment, adopting one or more of those approaches would require the definition of an orchestrator element to dispatch instances in a particular set of hypervisor hosts. At the heart of the launch process of instances is the nova-scheduler service, which determines on which host an instance will be placed. As highlighted in Chapters 1 and 3, the latest OpenStack releases, come with the Placement service, which was introduced in the Newton release. The Placement service works in tandem with the Nova scheduler for a more advanced and granular host pre-filtering selection process. To make the most of the segregation approaches defined in the previous section, it is essential to skim the surface of different compute scheduling...