60a87bad63
Now that nova supports libvirt's vDPA implementation [1][2], we need
deployment code to enable such a feature.
The ComputeVdpa role is basically a copy of ComputeSriov with the
addition of the ExtraKernelModules. Beside these modules, operators
should define to their ``NovaPCIPassthrough`` as they normally do
for standard SR-IOV as a role parameter.
When having the right modules loaded, at least in RHEL8.5, the kernel
will automatically create /dev/vhost-vdpa-X devices for each one of the
VFs.
[1] https://opendev.org/openstack/nova-specs/src/branch/master/specs/wallaby/implemented/libvirt-vdpa-support.rst
[2] https://bugzilla.redhat.com/1873409
Depends-On: https://review.opendev.org/c/openstack/os-net-config/+/801255/
Related: https://bugzilla.redhat.com/1917762
Change-Id: Ifda3ac08c596f65bc75d869097a0f72e816ba783
(cherry picked from commit 7010a27415)
12 lines
458 B
YAML
12 lines
458 B
YAML
---
|
|
features:
|
|
- |
|
|
The libvirt driver has added support for hardware-offloaded OVS
|
|
with vDPA (vhost Data Path Acceleration) type interfaces.
|
|
vDPA allows virtio net interfaces to be presented to the guest while
|
|
the datapath can be offloaded to a software or hardware implementation.
|
|
This enables high performance networking with the portablity of standard
|
|
virtio interfaces.
|
|
|
|
Nova added support for vhost-vdpa devices in Wallaby.
|