This feature adds support for PCI device NUMA affinity policies. These are configured as part of the `[pci]alias` configuration options. There are three policies supported: - `required` - `legacy` - `preferred` In all cases, strict NUMA affinity is provided if possible. The key difference between the policies is how much NUMA affinity you can forsake before failing to schedule. These policies allow you to configure how strict your NUMA affinity is on a per-device basis or, more specifically, per device alias. This is useful to ensure maximum resource utilization. When the 'preferred' policy is configured for a PCI device, nova now utilizes CPUs on a different NUMA node from the NUMA node of the PCI device if this is all that is available. This results in increased resource utilization with the downside of reduced performance for these instances.