Hyper-V: enable instance vNUMA

Registered by Claudiu Belu

Windows Hyper-V / Server 2012 introduces support for vNUMA topology into Hyper-V virtual machines. This feature improves the performance for VMs configured with large amounts of memory.

Instances will have to be enabled to use vNUMA, by setting the VirtualNumaEnabled flag to True in the Msvm_VirtualSystemSettingData object associated to them. See the source [1].

A few important notes:
* Host NUMA spanning is enabled by default.
* vNUMA and Dynamic Memory cannot be used at the same time.

For more details, check the source [2].

[1] http://msdn.microsoft.com/en-us/library/hh850257%28v=vs.85%29.aspx
[2] http://technet.microsoft.com/en-us/library/dn282282.aspx

Blueprint information

Status:
Complete
Approver:
John Garbutt
Priority:
Low
Drafter:
Claudiu Belu
Direction:
Approved
Assignee:
Claudiu Belu
Definition:
Approved
Series goal:
Accepted for ocata
Implementation:
Implemented
Milestone target:
milestone icon ocata-3
Started by
Claudiu Belu
Completed by
Matt Riedemann

Related branches

Sprints

Whiteboard

I think this needs to have a spec created. The configuration of NUMA topology in the guest is only one small part of the NUMA problem. The bigger thing is the host side accounting and scheduling of NUMA resources, and it isn't clear from those MSDN links how this might end up being done for hyperv in Nova. So I'd like see an expanded proposal in a spec for this. -- danpb 22/6/2015

Gerrit topic: https://review.openstack.org/#q,topic:bp/s,n,z

Addressed by: https://review.openstack.org/196162
    Adds Hyper-V vNUMA enable spec

Please note this blueprint will delayed until the M release if it is not in the NeedsCodeReview state (with all the code up for review) before July 16th, and merged by July 30th. We expect to re-open master for the M release in September. For more information, please see: https://wiki.openstack.org/wiki/Nova/Liberty_Release_Schedule#Non-priority_Blueprint_Feature_Freeze and http://lists.openstack.org/pipermail/openstack-dev/2015-June/065819.html
--johnthetubaguy 15th July 2015

Unapproved for liberty due to the Non-Priority Feature Proposal Freeze. --johnthetubaguy 16th July 2015

Gerrit topic: https://review.openstack.org/#q,topic:bp/hyperv-storage-qos,n,z

Addressed by: https://review.openstack.org/231417
    Moves the Hyper-V NUMA instance spec to Mitaka

Gerrit topic: https://review.openstack.org/#q,topic:bp/hyper-v-vnuma-enable,n,z

Reappvoed for mitaka --johnthetubaguy 7th October 2015

Pending Patches
==============

As there is no code up for review, removing this blueprint from mitaka, as part of the non-priority Feature Freeze. For more details please see: http://docs.openstack.org/developer/nova/process.html --johnthetubaguy 2015.01.15

Gerrit topic: https://review.openstack.org/#q,topic:bp/hyper-v-uefi-secureboot,n,z

Addressed by: https://review.openstack.org/279044
    Moves the Hyper-V NUMA instance spec to Newton

Addressed by: https://review.openstack.org/282407
    Hyper-V: Adds vNUMA implementation

We're past the non-priority feature freeze for Newton. This will have to be re-proposed for Ocata. I'd also recommend getting someone more familiar with NUMA in the libvirt driver, like sfinucan, danpb, cfriesen, etc to look at this when blueprints open up again for review in Ocata. -- mriedem 20160706

Addressed by: https://review.openstack.org/373483
    Reproposes the Hyper-V NUMA instance spec to Ocata

Addressed by: https://review.openstack.org/424116
    hyper-v: Do not allow instances with pinned CPUs to spawn

(?)

Work Items

This blueprint contains Public information 
Everyone can see this information.

Subscribers

No subscribers.