Open Stack

Excerpts from Hinds, Luke (Nokia - GB/Bristol)'s message of 2016-01-29 01:35:38 -0800:
> From: EXT Tomas Vondra [vondra at czech-itc.cz]
> Sent: Friday, January 29, 2016 9:04 AM
> To: openstack at lists.openstack.org
> Subject: Re: [Openstack] Cells: *how* experimental?
> 
> Clint Byrum <clint at ...> writes:
> 
> > > However, if you have some requirement to have everything under that
> > > one region, I can say that even in a 1000 hypervisor simulation I don't
> > > see more than 100Mbit of traffic to the control plane that all of the
> > > nodes share. I'd expect 30 nodes to be quite a bit less traffic.
> 
> > Hmm, simulation you say? What do you use to simulate an OpenStack?
> > Tomas
> 
<moved top post>
> I am also genuinely intrigued about this. Are your test results publically available Clint? 

We will be publishing our results, definitely, though I'm not sure
where or when, hopefully soon. The experiments are ongoing. I will be
submitting a talk to the next summit to present them as well, so we can
hope it is accepted too.

Basically we've spun up 1000 docker containers with the nova "fake"
virt driver, which doesn't actually start vms or plumb networks, but
lies to the control plane that it has done so. Then we're slamming that
with many client threads and seeing what it does to the control plane,
and what effect various configuration changes have on it.

Ideally we also land a smaller scale of this simulation as a test in the
openstack CI system so we can maintain progress on improving handling
such scale, similarly to the way the largeops job has prevented breaking
larger scale operations.

Stay tuned, I'll try to remember to reply to this thread when we
publish.