Ask Your Question

Openstack autonomic behaviour

asked 2014-09-08 04:40:14 -0500

affo gravatar image

Hi everybody! I already posted a similar question here.
But I felt like the question was not understood.
So, here it is, a new and better-explained question:

The context is:
OpenStack is a IaaS and everyone can launch instances and deploy applications on them.
Let's suppose we have 3 users and 3 apps on 3 VMs:
* user1 -> vm1 -> app1
* user2 -> vm2 -> app2
* user3 -> vm3 -> app3

Any of these users can setup an orchestration template on Heat and set alarms (thanks to Ceilometer) to perform automatic decisions in the context of the application and his/her VMs.

If all what I said is right, the question is:
Is there something like Heat in the underneath layer?
And so, is there something which orchestrates OpenStack itself and not the applications of end-users?

I mean, if I launch OpensStack on some servers, maybe I could be interested in setting some policies when something crucial happens.
For instance: when the number of VMs on this compute node is too high, let's switch to another compute node... Ans so on.

I hope that the question is clearer.
Thank you in advance.

edit retag flag offensive close merge delete


So, it is Nova which takes care of scaling horizontally on more compute nodes basing on the available resources?

affo gravatar imageaffo ( 2014-09-08 08:42:11 -0500 )edit

1 answer

Sort by ยป oldest newest most voted

answered 2014-09-08 08:22:39 -0500

Remko Duppen gravatar image

Nova-scheduler picks a hypervisor based on the request and available resources, as such it's inpossible to have an 'overdraft' and isn't it necessary to migrate vm's to another hypervisor.
Unless you do a resize, in which case nova will migrate the VM to another hypervisor if needed.

edit flag offensive delete link more


So, it is Nova which takes care of scaling horizontally on more compute nodes basing on the available resources? And, why do you say "it is impossible to have an overdraft"?

Anyway, can you suggest me a point from which to start investigating better on this point? Nova-scheduler doc maybe? Thank u

affo gravatar imageaffo ( 2014-09-09 05:42:37 -0500 )edit

The nova scheduler doesn't handle 'dynamic' scaling, it just picks a hypervisor with the resources available. If it can't find a suitable hypervisor it will throw an error. With overdraft I mean it's impossible' to claim more resources then present in your stack.

Remko Duppen gravatar imageRemko Duppen ( 2014-09-09 07:10:49 -0500 )edit

Yes, Nova takes care of the horizontal scaling, it's just not really dynamic

Remko Duppen gravatar imageRemko Duppen ( 2014-09-09 07:12:35 -0500 )edit

Ok, thank you. I will deepen my knowledge about this

affo gravatar imageaffo ( 2014-09-09 07:49:01 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower


Asked: 2014-09-08 04:40:14 -0500

Seen: 181 times

Last updated: Sep 08 '14