Bug 693187

Summary:	Test case live migration speed/performance impact.
Product:	Red Hat Enterprise Linux 6	Reporter:	Dor Laor <dlaor>
Component:	TestPlan-KVM	Assignee:	Mike Cao <bcao>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Keqin Hong <khong>
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	6.1	CC:	bcao, jskeoch, juzhang, llim, mdoyle, michen, tburke
Target Milestone:	rc	Keywords:	Tracking
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2011-04-20 12:11:46 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Dor Laor 2011-04-03 12:39:46 UTC

As can seen in https://bugzilla.redhat.com/show_bug.cgi?id=690521 we need to test the network bandwidth implications when doing live migration. In addition I like to see that we test also the performance over head for the guest when going through live migration.

Scenario: Let's run a guest with some app in it. The app can/should be one of cpu/net/blk intensive. Need to measure its performance before/during/post live migration. Performance == throughput and latency. We also need to measure the length of the downtime at the last stage of the live migration process.

Lastly we need to see that the live migration stage should converge assuming reasonable dirty page/sec number. This means that if the bandwidth for live migration is high enough (let's assume 1gb/s) migration finishes w/ donwtime of 0.1s - 0.5s.

A very nice and full data can be retried form the live migration bible document: http://www.cl.cam.ac.uk/research/srg/netos/papers/2005-migration-nsdi-pre.pdf

Comment 2 RHEL Program Management 2011-04-03 12:57:30 UTC

This request was evaluated by Red Hat Product Management for
inclusion in the current release of Red Hat Enterprise Linux.
Because the affected component is not scheduled to be updated
in the current release, Red Hat is unfortunately unable to
address this request at this time. Red Hat invites you to
ask your support representative to propose this request, if
appropriate and relevant, in the next release of Red Hat
Enterprise Linux. If you would like it considered as an
exception in the current release, please ask your support
representative.

Comment 3 Michael Doyle 2011-04-03 23:23:31 UTC

Mike, please let us know what changes are required to the KVM IEEE Test Plan to capture this testing.

Comment 4 Mike Cao 2011-04-07 11:46:56 UTC

16. Measure migration speed
-       migraion successed.After migration ,guest works fine.
+	Run a guest with some app in it. The app can/should be one of cpu/net/blk +intensive. Need to measure its performance before/during/post live
+ migration. Performance == throughput and latency migraion successed.
+       enlarge the migration speed during migration ,should check the +transferring speed is more or leass same as the speed set in qemu-monitor.
+migration. and migration should successed.
+        After migration ,guest works fine.

Comment 5 Mike Cao 2011-04-07 11:51:41 UTC

> Lastly we need to see that the live migration stage should converge assuming
> reasonable dirty page/sec number. This means that if the bandwidth for live
> migration is high enough (let's assume 1gb/s) migration finishes w/ donwtime of
> 0.1s - 0.5s.
Hi, Dor 

the default migration downtime is 0.3s ,in the testing ,We want to use ping command to measure it .but the offset of ping are much larger than migration max downtime. 
Could you provided me How to measure migration_max_downtime?

Thanks,
Mike 
> 
> A very nice and full data can be retried form the live migration bible
> document:
> http://www.cl.cam.ac.uk/research/srg/netos/papers/2005-migration-nsdi-pre.pdf

Comment 6 Mike Cao 2011-04-20 12:11:46 UTC

QE tried this case in RC migraion functional testing .did not find the bug described in comment #0.

I will close this issue.

Thanks,
Mike