Hi Mark,
just to be sure I got correctly your network setup - you want to stress test
full OpenStack installation by simulation of NIC and storage utilization, am I
right?
In that case vsperf can be used in "trafficgen mode", where it only controls
selected traffic generator and collects final results. As of now, vsperf
doesn't collect any periodic metrics from the traffic generator. It runs
selected traffic pattern (e.g. RFC2544 Throughput) and records measured values.
However I think that both scenarios would be possible, but first scenario would
have to be slightly changed:
1) I start up VSPERF on the NICs that represent the tenant network in an
OpenStack installation.
MK=> and measure e.g. throughput of setup without the storage network stressing
2) After 10 minutes, start up StorPerf read/write test that should stress
the OpenStack storage network
3) See if there is a change to the VSPERF throughput.
MK=> i.e. execute vsperf again with the same settings as before and
compare measured values
Second scenario can be executed as proposed by you, without any changes.
Best Regards,
Martin
From: [email protected]
[mailto:[email protected]] On Behalf Of Beierl, Mark
Sent: Thursday, August 17, 2017 4:39 PM
To: Yuyang (Gabriel) <[email protected]>
Cc: [email protected]
Subject: Re: [opnfv-tech-discuss] [VSPERF] [Bottlenecks] VSPERF and StorPerf
Hello, Gabriel.
The idea behind starting one first is so we can see the delta at the point
where it started. We certainly can run both independently as a baseline, and
then run them together can measure the delta that way...
Regards,
Mark
Mark Beierl
SW System Sr Principal Engineer
Dell EMC | Office of the CTO
mobile +1 613 314 8106<tel:1-613-314-8106>
[email protected]<mailto:[email protected]>
On Aug 17, 2017, at 04:20, Yuyang (Gabriel)
<[email protected]<mailto:[email protected]>> wrote:
Hi Mark,
I fully support the idea here. I think it is time testing projects together
develop some sophisticated stress test cases across components of the system.
Just 1 question, why start VSPERF and STORPERF separately. Why not start both
tests simultaneously and monitoring/reporting the throughput variation along
time.
Thanks,
Gabriel
From:
[email protected]<mailto:[email protected]>
[mailto:[email protected]] On Behalf Of Beierl, Mark
Sent: Thursday, August 17, 2017 12:17 AM
To:
[email protected]<mailto:[email protected]>
Subject: [opnfv-tech-discuss] [VSPERF] [Bottlenecks] VSPERF and StorPerf
Hello, Trevor and VSPERF team.
I'd like to get a quick overview of VSPERF and how the metrics are captured.
Sorry if these are dumb questions :)
First, is there a test that simply pushes packets over the network to see what
the maximum throughput is, like a saturation test?
Does VSPERF capture periodic metrics, or just a summary report at the end? If
I change something in the network while the test is running (such as adding
more traffic outside of VSPERF), is there a way to see the impact of that while
the test is running?
The reason behind the questions are for the following scenario:
1) I start up VSPERF on the NICs that represent the tenant network in an
OpenStack installation.
2) After 10 minutes, start up StorPerf read/write test that should stress the
OpenStack storage network
3) See if there is a change to the VSPERF throughput.
Alternatively, as StorPerf does have minute by minute metrics, we can do this:
1) Start StorPerf read/write
2) After 10 minutes, start VSPERF to stress the tenant network NICs
3) Look at the change in StorPerf throughput at the time that VSPERF starts.
Thoughts?
Regards,
Mark
Mark Beierl
SW System Sr Principal Engineer
Dell EMC | Office of the CTO
mobile +1 613 314 8106<tel:1-613-314-8106>
[email protected]<mailto:[email protected]>
--------------------------------------------------------------
Intel Research and Development Ireland Limited
Registered in Ireland
Registered Office: Collinstown Industrial Park, Leixlip, County Kildare
Registered Number: 308263
This e-mail and any attachments may contain confidential material for the sole
use of the intended recipient(s). Any review or distribution by others is
strictly prohibited. If you are not the intended recipient, please contact the
sender and delete all copies.
_______________________________________________
opnfv-tech-discuss mailing list
[email protected]
https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss
