How many daemons for openstack

I don't think you want multiple copies of the same daemon on the same machine right now.

You can have as many nova-api as you want, nova-scheduler is similar. They just pick messages of the same single queue as each decides to fetch the messages.

nova-network depends. You probably want HA mode with Flat DHCP, where you get a nova-network on every compute node, and the DHCP servers are configured such that they don't conflict.

nova-volume, I am not sure about. I think you have to set up storage zones, or something like that.

Can you make sure the docs cover this kind of thing, and if not, raise a bug?
http://www.referencearchitecture.org/physical-deployment/
http://docs.openstack.org/trunk/openstack-compute/admin/content/managing-volumes.html (etc...)

Thanks John for the answer.

So we can have multiple nova-api and nova-scheduler, now my question is: once customer submit request, which nova-api or nova-scheduler will do the job to handle the request?

Thanks!

I would expect nova api to behind a load-balancer, or virtual ip address, so it would be up to how that is setup.

The scheduler just takes messages off the queue when it feels it is ready, so it is whichever scheduler gets there first. At least that is how I understand it works. They all shared the same database, so it should not really matter who servers the request.

Thanks John for the help! Clear now.