14:59:34 #startmeeting advisorygroup 14:59:34 Meeting started Thu Sep 17 14:59:34 2015 UTC. The chair is phrobb. Information about MeetBot at http://ci.openstack.org/meetbot.html. 14:59:34 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:59:34 The meeting name has been set to 'advisorygroup' 15:00:36 #info Chris Luke is in the house 15:02:18 #info Jay Etchings 15:02:37 #info Pedro Gutierrez 15:02:44 #info Jamil Cawki 15:05:46 #info Alex Zhang 15:05:53 #info Chris Donley 15:05:59 #info Wade Shao 15:06:03 #info Jon Beasley 15:06:57 #info Chris Luke introduces the idea of independent, small work groups within the Advisory Group. 15:08:33 #info Chris Luke also notes that creating a way to have interactions with the TSC is also something the Advisory group should work toward 15:09:06 #info Zlex Zhang notes that the group may want to work on Use Cases from their companies as well 15:10:22 #info Ralf Trezeciak 15:10:48 #info Todd introduces Jay Echings to describe their experiences with ODL 15:12:58 #info Mr. Echings spoke about his first work with ODL and Internet 2. 15:16:48 #info Dr. Gail-Joon Ahn describes their work with FlowGuard and secure SDN models 15:21:50 #info within the next couple of weeks they will re-present their work as a funding opportunity... they also are partnering with Brocade although their goal is to be as vendor agnositc as possible. 15:22:10 #info they are looking to aggregage this as a unified ODL controller in production. 15:22:27 #info they now have their 2nd 100GB link to Internet2 15:23:12 #info The unification as it is defined is around ODL and the "open" methodology... they started with Cumulus but then realized they wouldn't have the flexibility they need 15:23:39 #info question - What was the problem with the cumulus networks? 15:25:31 #info Answer - the group working with Cumulus had thought it was an openflow compliant platform, but it wasn't fully. ASU maintained some of the equipment.. they had thought since they can boot different SW on top of the HW, they would have the flexibility they needed, but it did not turn out that way. 15:25:47 #info Q: regarding OpenFlow, was it used extensively? 15:26:49 my hotel connection is pretty bad, will read IRC -:) 15:27:17 #info Answwer - They had started with DANSES (defined apps for Science....) Jay and team wanted to have an unrestricted 40GB link they analyze the speed/latency of traffic. Using OF1.1 at the time... to Gail... 15:29:03 #info Gail notes OF with FlowGuard can provide more effective monitoring of the networking and the flowpath. Also even though part of work supporeted by DOE when they share their data.. whenever new policies are introduced they need to manage the new parhs needed. 15:33:39 #info Todd asked for feedback regarding specific topics for subgroups going forward 15:34:45 #info A: Ways to better quantify the uses of the controller. 15:38:14 #info Colin Dixon believes that the technical community does not currently have the clarity on the uses of ODL. We need to improve the use cases and the Documentation to make the content easier to consume. 15:38:27 #info ColinDixon notes that use cases are great and the tech. community hasn't had enough of them.. Related to that is documentation... either how it doesn't work or how it does work... those are excellent things to provide to the TSC. Either full gap analysis of what is missing, or what seems like it should be there but does not behave as expected are handy 15:39:25 #info Harvey Newman 15:39:27 #info ColinDixon notes that he is also the PTL for the Documentation project for ODL and he would be happy to help AG members to work on documentation 15:41:16 #info Alex Zhang asks that if there is anything on the technical side that can be brought to the Advisory Group to present ideas for Use Cases that members from the Advisory Group could focus on. 15:41:40 #info Alex Zhang notes by looking at release process and priorities, looking at the tech side if there are questions on the importance of a given feature or attribute of a feature, they TSC can come to the AG to ask for their opinion as well. For example, the NBI it is general and unclear.. a few initiatives to do intent-based model. It is not clear as to what is the winner there for the AG members to go work on. Having 15:41:41 discussions between the AG and the TSC on what is needed there may be helpful 15:42:54 #info Maybe the AG could combine the differnet feedbacks from the different AG members as a preso to the TSC as a summary. Alex notes that raising the key common issues would be helpful 15:46:22 https://wiki.opendaylight.org/view/OpenDaylight_Advisory_Group:Main 15:46:33 public list: https://lists.opendaylight.org/mailman/listinfo/advisory-group 15:47:01 #link https://wiki.opendaylight.org/view/OpenDaylight_Advisory_Group:Main the advisory group meeting wiki page with minutes and presentations 15:47:27 #link https://lists.opendaylight.org/mailman/listinfo/advisory-group the public advisory group mailing list 15:47:45 phrobb: I don’t have access to the private one (which is fine), so I can’t give the link 15:51:05 The private Advisory Group mailing list is (not unexpectedly) advisory-group-private@lists.opendaylight.org 15:53:38 #link https://lists.opendaylight.org/mailman/listinfo/advisory-group-private <-- link to the private AG mailing list. This site is only accessible to actual AG members. (hence the private designation) 15:56:44 #info Harvey Newman notes they had a hydrogen prototype, then Helium the Lithium... the system is huge. Maven downloads 1500 components... people have varied experiences.... they want more control on what gets loaded.... knowing what needs to be loaded is hard. RE usecases, there are some components that work on on HW where others don't. I knowledgebase is needed to document this stuff otherwise people trip over these 15:56:44 issues repeatedly. We need lots more info on what is robuts, and mature at any given time. 15:57:03 phrobb: noted :p 15:57:40 phrobb: we’ll see how moving some projects into mature will help, release notes try to get some of this across, but could use to be standardized and better publicized 15:58:03 phrobb: I’m not sure how to address the issue of 1500 libraries, that’s a deeper conversation 15:58:16 colindixon agreed.... also, I would have thought Karaf would help the other problem 15:59:20 FYI, last month I talked about the mature/stable stuff with this group 16:03:40 #info harvey? says that their experience with OpenDaylight gives has three pieces of feedback (a) maving downloading ~1500 libararies is troubling and there’s no good ways to control what gets loaded and understand what you really need, (b) there’s not a good way to understand which projects are how mature/stable/ready, (c) the comptibility matrix between hardware and different parts of opendaylight is absent in most cas 16:04:20 #endmeeting