On Wed, Oct 24, 2012 at 1:18 PM, Ben Pfaff <[email protected]> wrote: > That's a new aspect I noticed after I sent my initial response. I > don't actually know whether tests pass for every manufacturer but one, > because I do not use or test any switches other than the Open vSwitch > software switch. I have never, in fact, used any OpenFlow > implementation other than Open vSwitch (including what it was called > before it was called Open vSwitch). Nicira and (at least Nicira's > part of) VMware only uses the Open vSwitch software switch. > > I find it much more interesting to hear that Open vSwitch does > something different from every other OpenFlow switch, than to hear > that Open vSwitch does not conform to OFTest's idea of how an OpenFlow > switch should behave. Is it in fact true that the tests in question > pass on every OpenFlow switch other than OVS? How many is "every > other", by the way?
Hi Ben, I didn't think you were criticizing me and I hope you don't think I was criticizing you. I've always been a big fan of the OVS work and it was definitely what OFTest was developed against in terms of implementation details. My reply was just an honest attempt to get more people looking into the implementation differences. That said, to your questions: the last two ONF plugfests have had a phase where everyone has run their gear against OFTest. I don't remember the exact numbers, but IIRC there were ~10 switches at the first plugfest and ~15 at the most recent one. I was probably a little sloppy in my hypothetical "if a test passes for every manufacturer but one, who's job is it to investigate why it doesn't pass?" -- each vendor was not passing a few tests, but IIRC, there wasn't any specific test that was failing on any substantial fraction of boxes. - Rob . _______________________________________________ openflow-discuss mailing list [email protected] https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
