15:00:34 <serverascode> #startmeeting operators_telco_nfv
15:00:35 <openstack> Meeting started Wed Jan 25 15:00:34 2017 UTC and is due to finish in 60 minutes.  The chair is serverascode. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:36 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:39 <openstack> The meeting name has been set to 'operators_telco_nfv'
15:00:50 <serverascode> hello :)
15:00:56 <uli-k> hello
15:01:09 <shintaro> hi
15:01:23 <ad_rien_> o/
15:01:41 <serverascode> I'll give it another minute or so
15:02:02 <serverascode> if you hav eanything to add to the agenda, please go ahead :)
15:02:14 <serverascode> #link https://etherpad.openstack.org/p/ops-telco-nfv-meeting-agenda
15:02:26 <jamemcc> Hi
15:03:09 <serverascode> #topic Weekly Meetings
15:03:25 <serverascode> Not sure if anyone has an opinion on this, or I could push it to the next meeting
15:03:39 <serverascode> it was mentioned last meeting that perhaps we should move to weekly meetings
15:03:52 <ad_rien_> cannot on my side
15:03:58 <ad_rien_> to be honest
15:04:08 <ad_rien_> One week I have to chair the massively distributed WG
15:04:15 <ad_rien_> and the other one, I attend this meeting
15:04:26 <ad_rien_> it becomes to costly for me to attend all Openstack meeting
15:04:27 <ad_rien_> s
15:04:28 <ad_rien_> sorry
15:04:32 <serverascode> yeah I would like to start attending the massively distributed group more often myself
15:04:33 <GeraldK> hi
15:04:44 <serverascode> hi GeraldK :)
15:04:48 <ad_rien_> but if you believe it is mandatory, I will read the minutes
15:04:50 <ad_rien_> go ahead
15:04:54 <serverascode> I'm happy with bi-weekly meetings myelf
15:05:03 <ad_rien_> just to let you know that I will be not able to do more right now
15:05:22 <serverascode> anyone else have any strong feelings about it? otherwise I will suggest we revisit in a couple months
15:05:57 <jamemcc> every 2 weeks is good for me
15:06:19 <serverascode> ok, I put up a reminder to revist in a while
15:06:35 <serverascode> #topic Boston Summit
15:07:07 <GeraldK> every 2 weeks would be okay for me. Anyway, I have a meeting clash at this time.
15:07:30 <serverascode> Over the last couple days there has been some discussion on doing a presentation and session on multi-site/multi-cloud
15:08:03 <serverascode> so if anyone is interested in that, let me know, as we will start to put together and abstract to submit by Feb 6th
15:08:17 <ad_rien_> So what is your idea right now ?
15:08:29 <ad_rien_> are there notes somewhere?
15:09:16 <serverascode> no notes yet, just joehuang and I were discussing it, and we wanted to chat with you ad_rien_ regarding the Massively Dist. team
15:09:41 <serverascode> I think overall it was just that there is a lot of confusion regarding multi-site/cloud and we should help to clear that up :)
15:09:42 <ad_rien_> ok so the idea on our side is to present some performance evaluations of the scenario 1
15:09:47 <shintaro> is it from OPNFV? sounds interesting although i do not have any experience on the topic.
15:09:57 <shintaro> I mean muti-site
15:10:24 * ad_rien_ is looking for links
15:10:28 <serverascode> I think OPNFV has a project related to multi-site
15:10:38 <serverascode> but I'm more talking about an overview of all the different methods
15:10:40 <ad_rien_> #link https://docs.google.com/presentation/d/1jJFZejZqgYDxu5FX4K8g3I5zQ87afnjYI4VSRSuCQ6U/edit#slide=id.p
15:11:05 <ad_rien_> so basically we identified the different scenarios that can be done right now with the vanilla openstack code.
15:11:16 <serverascode> multi-region, multi-site, related projects like kingbird, tricircle, connecting tenant networks in different clouds, massively distributed, fog, edge...etc etc :)
15:11:34 <GeraldK> cool, thanks, thats a nice summary...
15:11:45 <ad_rien_> We implemented a tool dedicated to perform experiments. We plan to submit a presentation regarding the scenario 1
15:11:55 <shintaro> looks interesting topic
15:11:56 <ad_rien_> This should be tagged by the performance team
15:12:12 <ad_rien_> as discussed with DinaBelova yesterday
15:12:57 <ad_rien_> We can probably do another presentation that aims to clarify what is behind NFV infrastructures/massively distributed/Fog/Edge infrastructures
15:13:07 <serverascode> yeah I agree ad_rien_
15:13:08 <ad_rien_> and what is the link w-r-t the openstack community
15:13:30 <ad_rien_> I think it might be valuable (in addition to see whether we should not merge all the different WGs at the end ;) )
15:13:49 <serverascode> :)
15:14:20 <serverascode> ok, if you are interested in participating in putting forward that presenation and session, let me know and I will make sure to let you know what we are up to
15:14:25 <serverascode> I will also report back next meeting
15:14:33 <ad_rien_> We can also try to identify what are the production platform that are already deployed
15:14:42 <ad_rien_> We know the NECTAR one and the CERN one
15:14:47 <ad_rien_> I'm sure there are others
15:15:20 <ad_rien_> ATT has a massively distributed clouds but they are using their own glue to orchestrate the different openstack deployment
15:15:25 <serverascode> right, existing deployments would be a good to include
15:16:08 <serverascode> any other ideas on presentation topics this group could perhaps put forward?
15:16:19 <uli-k> Maybe Joe knows more. I heard something he will engage in the massive distributed cloud activity.
15:16:46 * uli-k (sorry I had nipped out a few minutes....)
15:17:24 <serverascode> I will still put forward a NFV session for Boston
15:17:40 <ad_rien_> sounds good
15:18:07 <serverascode> to be more clear, a working group meeting for this group
15:18:12 <serverascode> at the boston summit
15:18:15 <ad_rien_> sounds good
15:18:26 <serverascode> is there anything specific we want to achieve in that meeting? or leave it more general?
15:18:27 <ad_rien_> will do the same for the massively distributed
15:18:37 <ad_rien_> I will ask to do not schedule the two sessions at the same time
15:18:44 <serverascode> good idea :)
15:18:47 <ad_rien_> Can you do the same serverascode please?
15:18:55 <serverascode> right will do
15:18:58 <ad_rien_> thanks
15:19:58 <serverascode> for previous sessions they are often quite full, but end up being more like an general informational session on the what the working group is
15:20:18 <serverascode> and it doesn't necessarily move us forward all that much, though may help to find additional participants
15:21:10 <serverascode> actually it looks like our next meeting is after the deadline for submissions
15:21:19 <serverascode> so I will have to submit it before next meeting
15:21:40 <jamemcc> Regarding what to acheive in Boston.  Seems there is interest from those attending the Summit/Meetup.  Maybe we can hold the session to explore/document the varieties of the attendees NFV use cases.  Maybe we start with some fo the conversation we've had here.
15:22:28 <serverascode> ok, that is interesting. Would make use of the varied people attending the session.
15:23:26 <jamemcc> Kind of 2 fold - if we voted by end of the session we'd get some #'s as to the types and all in all might encourage those with the use cases that interest them to participate.
15:24:09 <jamemcc> What was the approcah in Barcelona Curtis - I was there but forgetting.  How did you kick it off?
15:24:23 <serverascode> Barcelona was pretty random
15:24:42 <serverascode> it was a small room, lots of people, and mostly it ended up being a discussion about what the working group actually is
15:24:59 <serverascode> perhaps people came out understanding what we are doing, but that was about it :)
15:25:34 <shintaro> 40min was a little too short I guess
15:25:41 <serverascode> yeah it is very very short
15:25:55 <ad_rien_> you can ask two consecutive slots?
15:26:03 <ad_rien_> we ask that for the massively distributed
15:26:09 <ad_rien_> it was enough I think
15:26:36 <serverascode> yeah that's an interesting idea, usually they are pretty limited on slots though, but yeah 40 min is so little
15:26:56 <serverascode> I will give it a shot :)
15:27:11 <GeraldK> in Barcelona we had collected many topics during that session but failed to attract people to IRC meetings to further discuss on them
15:28:05 <GeraldK> two slots could help to categorize/priorituze topics and also discuss next steps
15:28:12 <serverascode> right I am probably oversimplifying what we actually discussed, I should find the etherpad...
15:28:29 <serverascode> #link https://etherpad.openstack.org/p/BCN-ops-telcom-nfv-team
15:28:41 <ad_rien_> the other good point of having two consecutive slots is that you get people that are really interested to take part to the effort
15:29:00 <ad_rien_> i.e. folks that are not interested usually leave the room after 40 mins
15:29:39 <serverascode> is there anything in that etherpad that we could or should focus on for the session?
15:30:55 <serverascode> well, we can certainly alter the focus of the session later, now that I think about it the abstract doesn't have to be super specific
15:31:24 <GeraldK> serverascode: +1
15:32:09 <serverascode> ok next topic
15:32:20 <serverascode> # Generic NFV Platform Defintion Document
15:32:42 <serverascode> I started a bit of work on this in terms of starting a table of contents
15:32:45 <serverascode> #link https://etherpad.openstack.org/p/generic-nfv-platform
15:33:17 <serverascode> with the idea being of starting the document in that etherpad, filling it out as a first draft, then getting it into the git repo
15:33:31 <ad_rien_> +1
15:33:33 <ad_rien_> good work
15:34:05 <serverascode> thanks, it's really just an initial skeleton that needs a lot more work :)
15:34:22 <serverascode> but the etherpad is there, and if you have suggesions now or in the future, please feel free to make them :)
15:34:24 <ad_rien_> yes but at least it gives one direction
15:34:47 <jamemcc> It does look like a good outline - this is how we would would document the Reference Architecture?
15:35:38 <serverascode> I think so, my idea was to cover some NFV related background, then define an example architecture in one of the later sections
15:35:58 <serverascode> and that reference arch would be fairly prescriptive, but would juts be one example
15:36:18 <ad_rien_> I don't know whether it will be reachable but I think you have a good outline for the presentation ;)
15:37:53 <serverascode> yeah it will be tough to keep the scope limited enough to actually finish it and provide something useful
15:38:47 <serverascode> I see some additions already :) thanks!
15:39:04 <GeraldK> so, in setion 4 you'd like to collect basic features that should be there for a generic NFV platform?
15:39:53 <GeraldK> can we add a kind of Note on each section to explain what it should be about?
15:39:57 <serverascode> yup, or at least list the ones that are available in production
15:40:11 <GeraldK> do we need "advanced NFV features" for a generic NFV platform?
15:40:43 <serverascode> good point, I was thinking that we might have to at least mention them as I find that is where a lot of confusion comes int
15:40:58 <serverascode> or perhaps some way of determining what is "production" worthy
15:41:31 <GeraldK> so this is to differentiate what is "generic/basic NFV features" and "additional advanced featues"?
15:42:06 <ad_rien_> just one question (please take into account that I'm not involved in the OPENNFV community)
15:42:07 <serverascode> right, perhaps 'advanced' is the wrong word
15:42:16 <ad_rien_> Is there any overlappings with OPNFV actions
15:42:29 <ad_rien_> It looks to be general (i.e; interesting but general)
15:42:35 <ad_rien_> The focus should not be done on the OpenStack ?
15:43:18 <GeraldK> ad_rien_: I'd say so. however, for OPNFV I have not seen such a structured approach like serverascode has started here
15:43:28 <ad_rien_> ok
15:43:57 <GeraldK> and OPNFV is more than OpenStack; however, including SDN in the list, this might also go beyond here
15:44:00 <shintaro> but we can pick stuff from OPNFV requirements
15:44:07 <ad_rien_> one more question
15:44:22 <ad_rien_> What about having a use-case that can guide the discussion
15:44:30 <GeraldK> shintaro: certainly yes. I'd even love to see both work together on such a document.
15:44:35 <ad_rien_> (i.e. don't know; cloudRan for instance if it makes sense)
15:44:54 <serverascode> a use case might be a good idea
15:45:25 <serverascode> for the SDN section, I think it's valuable to discuss b/c at some point basic neutron might have all the features one would require from an SDN :)
15:45:46 <serverascode> I mean if if gets service function chaining... :)
15:45:51 <GeraldK> this is something I also had in mind with the previous topic and the long list of interesting topics from last Summit: let's try to see which topics could be well described by a user story and let's propose such user story in the PWG if it does not yet exist
15:46:31 <GeraldK> serverascode: you're right
15:48:11 <shintaro> serverascode VNF can come in containers as well or may need to connect with PNF(physicalNF) so it might go beyond Neutron
15:48:27 <GeraldK> about "3. commen theme use case": do you have a use case in mind why we need a generic NFV platform or a use case how it could be used or something different?
15:49:05 <serverascode> shintaro good point, I haven't even begun to consider containers in NFV yet but they are important
15:49:11 <ad_rien_> The idea is to introduce a use-case that will act as a common theme for all following aspects
15:49:34 <ad_rien_> This should make the read and the understanding easier (I'm not sure we can be exhaustive i.e. covering all use-cases)
15:49:37 <ad_rien_> I can be wrong ;)
15:50:45 <serverascode> some kind of VNF that has clustering built into the application would probably be the baseline
15:50:53 <serverascode> not sure what that would be
15:51:09 <serverascode> I find lots of telecoms struggle with the lack of "instance HA" in openstack
15:51:37 <serverascode> they want to run single isolated workloads and have them magically be restarted if the hypervisor fails
15:52:05 <serverascode> I know there are some projects in the big tent to do that, but generally speaking that is not a usecase openstack supports
15:53:14 <shintaro> yeah we are working on that project. but that is for "legacy"apps
15:53:21 <serverascode> ok, so we have about 7 minutes left, should we go into open discussion?
15:54:03 <shintaro> I want to make sure whether we will have a session/meetup in OpsMeetup in Milano
15:54:17 <serverascode> #topic open discussion
15:54:27 <serverascode> shintaro: good call
15:54:33 <shintaro> Session discussion is open and now is time to fix it
15:55:04 <serverascode> where is the discussion happening now? still in the etherpad?
15:55:30 <shintaro> #link https://etherpad.openstack.org/p/MIL-ops-meetup
15:55:46 <ad_rien_> I'm lost sorry
15:55:58 <ad_rien_> ok
15:56:02 <ad_rien_> I'm back
15:56:02 <ad_rien_> sorry
15:56:08 <serverascode> there is an operators meeting in Milan in March, and we might have a session there :)
15:56:13 <ad_rien_> taking about the Milano meetup
15:56:25 <shintaro> Registration is now open, too
15:56:28 <serverascode> I think some people from the LCOO will be there?
15:56:52 <shintaro> #link https://opsmidcycle_march2017.eventbrite.com
15:57:03 <serverascode> the NFV session discussion got a little confused when I think Chris was suggesting maybe it's not a session
15:57:14 <serverascode> but it depends on how many NFV related pepole will be there
15:57:19 <serverascode> usually there are only one or two
15:57:31 <serverascode> everyone else is not doing anything that they would consider NFV
15:57:54 <serverascode> oh but now I see some +1s :)
15:58:25 <shintaro> LCOO decided not to have an official WG meeting there but some members will be there, I guess.
15:58:53 <serverascode> so you and I will be there, but I'm wondering who else?
15:59:08 <serverascode> oh one min left :)
15:59:12 <ad_rien_> Not me
15:59:16 <serverascode> I will have to end the meeting soon
15:59:21 <ad_rien_> but if we can attend by a webconf tool it would be great
15:59:27 <ad_rien_> ;)
15:59:29 <serverascode> shintaro: if you have time we can continue the discussion by email or irc
15:59:54 <shintaro> ok, but email would be better since its 1am for me
16:00:00 <serverascode> if you want you can email me at curtis@serverascode.com
16:00:13 <serverascode> and we can discuss that way or setup a time
16:00:18 <serverascode> gotta end meeting!
16:00:21 <serverascode> #endmeeting