Optimize NUMA topology based deployments on PPC arch

Registered by Sudipta Biswas on 2015-11-20

PPC architecture does core based scheduling for processes.
Openstack's libvirt driver presently only does thread based cpu scheduling. The same libvirt driver is used in case of PPC based systems as well.

This has implications in case of NUMA cell affinity and dedicated CPU pinning in PPC.

This blueprint aims to solve the problem by ensuring that NUMA cell affinity and dedicated CPU pinning works optimally on PPC arch.

Blueprint information

Status:
Not started
Approver:
None
Priority:
Undefined
Drafter:
Sudipta Biswas
Direction:
Needs approval
Assignee:
Sudipta Biswas
Definition:
New
Series goal:
Proposed for mitaka
Implementation:
Unknown
Milestone target:
None

Related branches

Sprints

Whiteboard

I don't really understand the context from the description, I think we will need a very small spec to describe this. --johnthetubaguy 20151127

Gerrit topic: https://review.openstack.org/#q,topic:bp/numa-optimization-ppc,n,z

Addressed by: https://review.openstack.org/251296
    PowerPC architecture does core based scheduling for processes. Openstack's libvirt driver presently only does thread based cpu scheduling. The same libvirt driver is used in case of PowerPC based systems as well.

(?)

Work Items

This blueprint contains Public information 
Everyone can see this information.

Subscribers

No subscribers.