On Fri, Mar 6, 2015 at 2:09 PM, Ben Pfaff <b...@nicira.com> wrote: > On Fri, Mar 06, 2015 at 08:36:49AM -0800, Gurucharan Shetty wrote: >> The design was come up after inputs and discussions with multiple >> people, including (in alphabetical order) Aaron Rosen, Ben Pfaff, >> Ganesan Chandrashekhar, Justin Pettit, Russell Bryant and Somik Behera. >> This commit introduces some complications around the Bindings table >> of OVN schema that is yet to be sorted out. So this is a early version. >> >> Signed-off-by: Gurucharan Shetty <gshe...@nicira.com> > > I think that there are two different things going on here. One is > integration of containers with OVS, which seems like something that > IntegrationGuide.md should cover. This is a little tricky and I have tried to write something inside the IntegrationGuide.md for Containers for a while but was never satisfied with what it would eventually look like.
The focus of IntegrationGuide.md is on hypervisors. So far, IntegrationGuide.md covers everything that an integration with Containers would need, when they are run inside a hypervisor. Though it does not mention Containers anywhere, whatever is being done for a VM, the same thing can be done for Containers. But. running Containers on hypervisors is not currently considered safe. So, I was not sure what can be written there that would actually be useful to anyone. >The other is the further > integration of containers with OVN; this is a very good description. In this case, there is nothing written inside the OVSDB for context (other than VLANs). So again, I skipped IntegrationGuide.md > > I'm not sure of how you chose to split content between the new > document and ovn-architecture(7). I'm inclined to try to keep as much > of the OVN architecture in ovn-architecture(7). The proposal had 2 parts to it. 1. Integration of Containers with OVN. I added this information in ovn-architecture. This talks more about OVN tables and utilities. This would not be limited just to OpenStack. 2. Changes needed in OpenStack (it needs to handle two extra arguments during port create. It also needs to verify credentials against the passed vif-id) to handle the containers workflow (Since this is new to OpenStack, I felt that the readers could more easily relate with the workflow if shown from OpenStack perspective). So I created a new document. Probably the document name should be different? > > In ovn-architecture(7) it might be worth motivating why we use tags > instead of one VIF per container, because we've had that question a > few times and it's a reasonable one. I agree. I will see to incorporate this one in v3. > > I like the term CIF, thanks for inventing it. (At least I haven't > seen it before.) I don't know who invented it! It could be Justin or Somik (or me). > > I need to think about this proposal a bit more. I'll do that over the > weekend. Thanks! I am sure you will come up with a better idea! _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
