Bug 657086

Summary:	[RFE] Limit number of VMs being migrated at the same time according to bandwidth
Product:	[oVirt] vdsm	Reporter:	Dan Kenigsberg <danken>
Component:	RFEs	Assignee:	Doron Fediuck <dfediuck>
Status:	CLOSED WONTFIX	QA Contact:	Nikolai Sednev <nsednev>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	---	CC:	aflierl, bazulay, bsettle, bugs, dfediuck, dmoessne, lpeer, mgoldboi, michal.skrivanek, mkalinin, mkenneth, nyechiel, pbatkowski, rbalakri, Rhev-m-bugs, sherold, sputhenp, srevivo, tao
Target Milestone:	---	Keywords:	FutureFeature, Triaged
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Enhancement
Doc Text:		Story Points:	---
Clone Of:	623713
Clones:	657089 (view as bug list)		Environment:
Last Closed:	2017-06-08 00:51:52 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	SLA	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:	623713, 696138, 1252426
Bug Blocks:

Comment 2 Barak 2010-11-29 07:32:10 UTC

Added conditional nak on 'design'
There are various factors that influence migration:
1 - VM's memory size
2 - bandwidth to destination
3 - source CPU load
4 - destination CPU load
5 - concurrent migrations (actually this one influence #2 & #3)

Since bandwidth is only one consideration among all the rest, and there can't be 1 migration policy that fits all,  we need to come out with few migration policies and let the administrator determine which one to use.

This needs design in both VDSM & RHEVM

Comment 5 Doron Fediuck 2014-01-06 15:39:32 UTC

Scott, please look into this issue and see if we can understand the needs
based on existing migration priority.

Comment 6 Scott Herold 2014-04-23 20:17:12 UTC

Once we implement Host-Level Network QoS (RHEV BZ 1043226), won't that provide a solution for this?

Comment 7 Nir Yechiel 2014-04-24 09:22:27 UTC

Yes, proper QoS per physical interface/VLAN device should provide a solution for this. There is also this one [1] which can be relevant here.

-Nir

[1] https://bugzilla.redhat.com/show_bug.cgi?id=1043809

Comment 8 Doron Fediuck 2014-04-30 12:24:12 UTC

Based on comment 7, pushing to 3.6, to revisit if needed.

Comment 9 Michal Skrivanek 2015-08-11 11:53:08 UTC

regardless QoS on the network we want to improve migration behavior in general aligned with comment #2, see bug 1252426

Comment 10 Michal Skrivanek 2015-08-11 11:59:17 UTC

setting bug 1252426 as Depends On, even though it doesn't really depend on that. QoS could be done independently. One can look at it as part of a general migration improvement, but I don't plan to track it under bug 1252426, so keeping this bug open for further consideration

Comment 15 Moran Goldboim 2017-06-08 00:51:52 UTC

Following live migration improvements happened on 4.0 and continued on 4.1, we currently don't see a need from the field to prioritize this functionality. please reopen if this still an issue.