Bug 1144606

Summary: [RFE] Implement virtio multiqueue usage in Libvirt driver
Product: Red Hat OpenStack Reporter: Stephen Gordon <sgordon>
Component: openstack-novaAssignee: Vladik Romanovsky <vromanov>
Status: CLOSED DUPLICATE QA Contact: Ami Jeain <ajeain>
Severity: high Docs Contact:
Priority: high    
Version: 7.0 (Kilo)CC: atheurer, ndipanov, yeylon
Target Milestone: gaKeywords: FutureFeature
Target Release: 8.0 (Liberty)   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Enhancement
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2015-03-13 19:30:18 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Stephen Gordon 2014-09-19 21:15:34 UTC
Description of problem:

Take advantage of virtio net multiqueue support by tuning number of queues automatically as the number of vCPUs increases:

    http://fedoraproject.org/wiki/Features/MQ_virtio_net

Version-Release number of selected component (if applicable):

Kilo

Comment 1 Stephen Gordon 2014-09-19 21:17:02 UTC
Andrew do we have any existing performance guidlines on exactly how to best scale this in relation to the number of vCPUs. We could expose this as a tunable flavor extras specification or image property but I'd prefer to rely on intelligent behaviour implemented in Nova if possible.

Comment 2 Stephen Gordon 2014-10-10 19:42:00 UTC
Discussing otp it seems like the number should be directly correlated to the # of vCPU cores, up to the maximum number of queues permitted?

Comment 3 Andrew Theurer 2014-10-20 14:20:00 UTC
Yes, I would recommend 1 queue per vcpu up to the limit (8 I believe).

Comment 4 Stephen Gordon 2015-03-13 19:30:18 UTC

*** This bug has been marked as a duplicate of bug 1153440 ***