Open Stack

Hi Akihiro

Thank you for your summarize this!

so We will support Hybrid driver in future?
Yong mentioned we should remove that driver if we have openflow based
ovs security group.
if so, it don't make sense if we have parameter for the function.





2013/2/8 Akihiro MOTOKI <amotoki at gmail.com>:
> Hi, (folks involving vif-pluginng improvements)
>
> In the review https://review.openstack.org/#/c/19126/,
> there is a dicussion whether we should address the issue
> about Hybrid VIF Driver with Quantum OVS security group in this patch.
>
> I summarized the relationship between VIF driver and Nova/Quantum
> security group implementaiton.
> https://docs.google.com/presentation/d/1Gro8rFJ8eYGx5u7BsHbv6WU4tq9v8chqAQSKwWzGUok/edit#slide=id.ga808ce59_00
>
> To address the issue about Hybrid driver with Quantum OVS security group,
> Quantum needs to pass two additional types of information:
> (1) the one is whether quantum provides firewalling
> (2) the other is whether hybrid vif-plugging (additional linux bridge)
> is required.
>
> I plan to add new attributes to Quantum binding extensions for a port:
> "binding:has_firewall" and "binding:bridge_required".
> Previously I introduced "binding:capabilities[port_filter]", but on a
> second thought
> it would be better to move it to the top level attribute since a sub-dict model
> for an attribute makes it difficult to validate it on the API layer.
>
> Any comments on the direction and the names of the attributes are also
> appreciated.
>
> It requires both Quantum and Nova updates, but these patches will be small.
>
> Thanks,
> Akihiro
>
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev