On Tue, Jun 14, 2011 at 08:03:38AM -0700, Peter Phaal wrote: > The porting guide shows an "ofproto provider" block in the first diagram, > which I take to be the ofproto plugin? This block is dropped in the second > diagram.
The block labeled ofproto-dpif in the second diagram is an example of an ofproto provider; that is, "ofproto provider" is the interface and "ofproto-dpif" is one implementation of it. (Is there a good way to illustrate that in the diagrams? I can see why it is confusing, but I don't yet see how to improve it.) > The vSwitch case is likely to see the most hardware diversity since > different vendor's network adapters might be installed in the same > server. Some adapters may have suitable hardware support for > OpenFlow and others won't. Making it clear what the driver boundary > is for the adapter makes it easier to see how they would all plug > into OVS as a common control plane. It looks like the ofproto > provider API abstracts the hardware differences in the forwarding > capabilities. Is this correct? That's the intention, yes. _______________________________________________ discuss mailing list discuss@openvswitch.org http://openvswitch.org/mailman/listinfo/discuss
