05:32:33 <soichi> #startmeeting taas
05:32:33 <anil_rao> Have we started the meeting?
05:32:33 <openstack> Meeting started Wed Oct 19 05:32:33 2016 UTC and is due to finish in 60 minutes.  The chair is soichi. Information about MeetBot at http://wiki.debian.org/MeetBot.
05:32:34 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
05:32:36 <openstack> The meeting name has been set to 'taas'
05:32:45 <anil_rao> Thanks! :-)
05:32:54 <soichi> #topic: Open Discussion
05:34:16 <soichi> vBrownBag Tech Talk in Barcelona Summit
05:34:36 <soichi> anil_rao: Thank you for the latest version of slides. It looks fine.
05:34:59 <anil_rao> soichi: Did the animations work correctly?
05:35:13 <soichi> yes
05:35:24 <anil_rao> Thanks for verifying
05:35:40 <kaz> I think it is easy to understand.
05:36:41 <anil_rao> Kaz: Thanks. It took quite some time to get that to work, especially since I am not very good with PowerPoint. :-)
05:37:14 <anil_rao> I see that Vinay (and others) will also be presenting a vBrownBag talk.
05:37:21 <soichi> yes
05:37:31 <soichi> Oct. 25th Tuesday 14:15-14:30 "Tapping in NFV cloud : A real world showcase by Swisscom, Ericsson and IXIA"
05:37:49 <anil_rao> It should be an interesting presentation.
05:37:57 <soichi> sure
05:39:29 <anil_rao> I think we should meet with the folks who are invested in the extension driver in Barcelona.
05:39:51 <soichi> +1
05:40:13 <kaz> +1
05:40:50 <anil_rao> Would you know if that work (i.e. the framework) is scheduled for the O release.
05:41:33 <soichi> I don't know. I guess some session in the Design Summit.
05:42:12 <anil_rao> OK.
05:44:38 <soichi> For the preparation of presentation, I'd like to propose the final check and a rehearsal at luch time on 26th Wednesday.
05:44:39 <soichi> I think time keeping is very important for the short time (10 minutes) presentation.
05:45:06 <anil_rao> +1
05:46:49 <soichi> I'm not sure, but i guess 3 min. for Reedip and 4 min. each for Anil and Kaz, for example.  11min. :-)
05:47:13 <anil_rao> That sounds good.
05:47:24 <soichi> thank you
05:49:17 <soichi> Let's finalize the slides on WebEx meeting, tomorrow.
05:49:37 <anil_rao> Sure. I'll set up that meeting for the same time as last week.
05:49:43 <anil_rao> One other thing...
05:49:50 <soichi> yes
05:50:49 <anil_rao> kaz,soichi: During the Summit I'd like to chat with you some more about the isolated tunnel case and we should try to get that in soon.
05:51:13 <soichi> +2
05:51:48 <soichi> I  hope Yamamoto can join us.
05:51:49 <anil_rao> It is a very useful feature, especially for folks who have an existing cloud infrastructure and want to enable traffic monitoring on it.
05:52:08 <anil_rao> Yes. We'll meat as a team and discuss this (and perhaps other items too).
05:52:25 <anil_rao> meat --> meet
05:52:43 <soichi> sure
05:53:43 <soichi> let's coordinate date. time, and place by email.
05:53:57 <anil_rao> I also want to talk about monitoring ports on the network nodes, which would be an interesting project in its own right.
05:54:19 <soichi> +1
05:55:30 <anil_rao> Another topic would be the items that we need to address in preparation for inclusion in the Big Tent / Neutron Stadium.
05:55:52 <anil_rao> There are several things that need to be handled when errors occur or things fail.
05:56:21 <soichi> I agree.
05:57:33 <soichi> inclusiton to the Neutorn Stadium should be a big topic for us
05:57:55 <anil_rao> Agreed
05:59:24 <soichi> we really want to make TaaS as one of the default functions (I mean TaaS is installed as one of the default functions of Neutron)
06:00:02 <yamamoto_> hi
06:00:11 <anil_rao> Hi
06:00:11 <soichi> yamamoto_: hi
06:00:25 <kaz> hi
06:00:40 <soichi> yamamoto_: i'm looking forward to meeting you in Barcelona
06:00:54 <anil_rao> +1
06:01:19 <anil_rao> I plan to look into the restartability issue with our TaaS Agent.
06:02:11 <anil_rao> I think this will be crucial for inclusion in the Neutron Stadium.
06:02:42 <soichi> i see
06:03:57 <yamamoto_> anil_rao: i guess it's one of our favorite topics.  "make it an l2 agent extension"
06:04:38 <anil_rao> +1
06:07:34 <yamamoto_> reedip owns the task. but it seems he is not here today?
06:07:57 <soichi> yamamoto_: yes
06:08:50 <yamamoto_> probably someone should step up and take it over from him.
06:11:13 <anil_rao> Let's discuss this during the Summit next week.
06:11:20 <soichi> Reedip will also come to Barcelona, so I think we can disscuss this topic there.
06:11:25 <soichi> anil_rao: +1
06:11:48 <anil_rao> Outside of just the agent itself, we also need some error correction login in the Plugin code that runs in the controller.
06:12:02 <yamamoto_> and/or maybe write some code there?
06:12:43 <anil_rao> login --> logic
06:13:13 <yamamoto_> anil_rao: do you have a bug for that?
06:13:25 <anil_rao> I'll file one
06:13:30 <yamamoto_> thank you
06:15:11 <yamamoto_> it reminded me LP ownership issue.
06:16:10 <yamamoto_> anil_rao: did you hear anything about it from vinay recently?
06:17:10 <anil_rao> yamamoto_: No. However, Vinay will be there too so we can discuss this in person with him.
06:17:57 <yamamoto_> is there anything to discuss?  i thought it was just "vinay doesn't have spare time to do it" type issue.
06:18:34 <anil_rao> Didn't know that.
06:19:28 <soichi> i guess he just forgot it.
06:19:54 <anil_rao> Can we discuss some performance related items?
06:20:10 <yamamoto_> sure
06:20:38 <anil_rao> Kaz and Soichi have given us a good start by collecting some very intersting data (for both the shared and isolated underlay cases).
06:21:06 <anil_rao> I would like to have this work continue. I'll pitch in to this effort with some experiments of my own.
06:21:35 <soichi> +1
06:21:39 <kaz> yes
06:22:05 <anil_rao> One of the things I want to evaluate is how much monitoring we can do on a given host before it is deemed too much. i.e. before it starts affecting other VMs on the host.
06:22:28 <soichi> agree
06:23:01 <anil_rao> We eventually should try to tie this up with the TaaS framework so that this data (perhaps some form of statistics) can be made avialable to the end-user.
06:23:40 <anil_rao> For example how much mirrored traffic is flowing on a per-host or per-tenant basis.
06:24:46 <soichi> i agree with you. And also on a per-tap-service, i think.
06:24:55 <anil_rao> Yes.
06:25:33 <yamamoto_> i guess neutron itself is weak in that area.
06:25:34 <anil_rao> Without this kind of statistics it is very hard  to know what is happening on a multi-host, multi-tenant system when a facility like distributed port-mirroring is available.
06:25:58 <anil_rao> yamamoto_: Agree
06:26:21 <soichi> anil_rao: agree
06:27:05 <anil_rao> TaaS can make this Neutron weakness worse because one can have several silent consumers of cpu and network bandwidth which are end-user initialted tap-service instances.
06:27:29 <yamamoto_> there was "physical topology" effort while ago but i don't know where it went.
06:28:30 <yamamoto_> anil_rao: valid concern
06:30:11 <anil_rao> If you all are interest I can discuss some ideas I have on this topic with you next week.
06:30:25 <soichi> sure
06:30:29 <kaz> yes
06:30:57 <anil_rao> I think we are out of time.
06:31:05 <soichi> yes, now it looks we have run out of time
06:31:07 <soichi> We will skip next week IRC because of Barcelona Summit.
06:31:15 <anil_rao> Sure
06:31:20 <yamamoto_> with midonet you can visualize taas traffic. (promotion mode :-)
06:31:34 <soichi> :)
06:31:37 <soichi> #endmeeting