ONF PlugFest: Day 2 (Indianapolis), Day 3 (Beijing)
Something about making history with SDN connections half a world away sounds really exciting.
Until you have to do it.
Lots of scrambling around today to get the testing set up for the SDN connections and testing later this week over the circuit. What kind of latency is acceptable over that kind of distance? No one is sure, but we are about to find out.
One of the other things that needs to get done at PlugFests is documentation—lots of it.
Documentation about where the OpenFlow® 1.3 specification may be ambiguous.
Documentation about how accurate a test case may be in illustrating a particular behavior.
Documentation about where one product does not work with another one.
Documentation. It’s something that just seems to be a bit pain in the rear when you have a lot of other things to do. In order to stimulate the authoring of this kind of documentation, I’ve taken to giving away digital trinkets and baubles for those teams that are excelling in what we need. So this morning, the team from Ixia walked into some free Google Chromecast devices for their hard work--we got lots of really vital intel that will help the entire industry as a result of their work. All of a sudden, that got the Luxoft and Spirent teams juiced to get some better work in time for the deadlines today.
The first group huddle-up between the 16 companies at the InCNTRE lab in Indianapolis and the 9 companies that are at BII (Beijing Internet Institute) came off well last night—of course the folks in China came in looking tired from an early morning start at 7am their time, but the 40 some folks who stayed late after a full first day of installs, configs, and initial pinging and testing looked pretty fried, too. Lots of good information flowing back and forth, and the video links worked pretty well, but all in all, both teams looked like a good night sleep is what they needed more than anything.
Today seems to have a lot more energy. Discussion flowing around the various live reporting that each of the companies is supposed to perform—how an application works, how a particular controller works with a particular switch, various versions of OF-Test, OF-Config, other testing frameworks—all of these interactions are detailed so that feedback to the OpenFlow® specification takes place, in addition to the normal bug fixes that each companies sees. I wish I could tell you all the “company A had problems with company B” kind of stuff, but it happens and it’s fixed almost before anyone can write it down. Such is the benefit of a rapid-development environment that has a blanket NDA umbrella to protect conversations and encourage collaborative bug-fixing among companies.
Several good meetings hashing through all the issues—what works well, what needs work, and which of the many features in OpenFlow® 1.3 should be included as mandatory in the conformance testing, and which should be regarded as vendor-specific and therefore optional. Great to read today that HP is making a significant contribution of code and FTE’s to the OpenDaylight Project—the HP guys here were gushing about how much it will be helping our SDN efforts. I’ll be talking with some of the HP folks and a few other engineers in some interviews that will be include in tomorrow’s blog.
If there is one thing that ONF tried to do at a plugfest, it's to keep everyone happy and well fed. The caffeine level is acceptably high—both cold and hot versions exist throughout the day and into the night—and the food service is excellent. The Indiana University food service people (college food 2.0 is a lot different from the "mystery meat" stuff we had in the dorms in the 1800's) do a great job making sure there are hot meals for lunch and a few late dinners (of course, the US group would be having Chinese food on the menu for dinner on the evening of the joint US/China SDN testing!). We’ll take a break to watch the local NBA team tonight, but hit it hard in the morning.
Rick Bauer - Technical Program Manager