9a2a1a375d
Adds some conceptual information about NUMA, and discusses how nova considers NUMA topology when booting instances. (This is a starting point, I'm happy for people to move this to a different section, or change wording). Change-Id: I328831fb3aef9531ee924aa86726277b7eac0cb5 Closes-Bug: 1369207
1.2 KiB
Consider NUMA topology when booting instances
NUMA topology can exist on both the physical hardware of the host, and the virtual hardware of the instance. OpenStack Compute uses libvirt to tune instances to take advantage of NUMA topologies. The libvirt driver boot process looks at the NUMA topology field of both the instance and the host it is being booted on, and uses that information to generate an appropriate configuration.
If the host is NUMA capable, but the instance has not requested a NUMA topology, Compute attempts to pack the instance into a single cell. If this fails, though, Compute will not continue to try.
If the host is NUMA capable, and the instance has requested a specific NUMA topology, Compute will try to pin the vCPUs of different NUMA cells on the instance to the corresponding NUMA cells on the host. It will also expose the NUMA topology of the instance to the guest OS.
If you want Compute to pin a particular vCPU as part of this process,
set the vcpu_pin_set parameter in the nova.conf configuration
file. For more information about the vcpu_pin_set
parameter, see the Configuration Reference Guide.