integrating with quantum - proposal

[Saggi Mizrahi]

This design is kind of odd.
You send the config to the quantum server.
Then you ask vdsm to tell the quantum agent to read the configuration from the server.

The Quantum server looks a bit redundant.
Ideally you would just send the conf to VSDM that will push it to the Quantom Agent it in turn will configure the DHCP.
Alternatively, Quantum Server pushes the information to the agents.

Now, from what I understand Quantum is "pull based" because it needs to be agnostic to VM allocation.
This means that the actual design should be that when running a VM part of the information is the openstack network ID.
VDSM in turn gets the agent to get the configuration info and configure the network.

It seems like the setDHCP() can be implied when starting a VM.
Also setDHCP() is kind of a bad name for this method as it doesn't set the protocl.
setDHCPConfigration() maybe but I'd rather have setIMAPConfiguration() so it's a bit less specific about how things are implemented.

In any case, the flows described in the wiki are part free text part flow diagram part some sort of flow\sequence diagram hybrid.
This makes it a bit difficult for me to follow the entire process. I might just have gotten everything wrong :).
Also, it's unclear to me who actually configures the network.

On 11/26/2012 09:06 AM, Livnat Peer wrote:
> Hi Guys,
> I'm about to send the proposal I worked on with Kolesnik to the
> list.
>
> Any political/other objections?
>
> http://wiki.ovirt.org/wiki/Quantum_IPAM_Integration
>
> We are starting to work on a POC to find the issues we missed in
> our
> research so far, sending this to the list can get the discussion
> going.
>
> Livnat
>