06:30:11 <vnyyad> #startmeeting taas
06:30:12 <openstack> Meeting started Wed Nov 18 06:30:11 2015 UTC and is due to finish in 60 minutes.  The chair is vnyyad. Information about MeetBot at http://wiki.debian.org/MeetBot.
06:30:13 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
06:30:15 <openstack> The meeting name has been set to 'taas'
06:30:17 <yamamoto> hi
06:30:20 <vnyyad> Hi
06:30:23 <soichi> hi
06:30:31 <kaz> Hello
06:30:50 <dedery> good morning/noon/afternoon/night :)
06:30:55 <reedip> o/
06:31:51 <vnyyad> shall we get into the agenda
06:32:22 <vnyyad> thanks soichi for proposing the items for this weeks agenda
06:32:33 <yamamoto> #link https://wiki.openstack.org/wiki/Meetings/taas Agenda
06:33:48 <soichi> i got several important comments from Anil.
06:34:43 <vnyyad> i guess anil is not in the meeting today but as he replied his commtnst to all of us we can discuss them
06:35:07 <yamamoto> soichi: kaz: are you sure that "A mirrored traffic double back to br_int" is a real disadvantage?
06:35:49 <soichi> yes, in terms of efficency.
06:35:54 <yamamoto> as all userland bridges connected by patch ports are handled by a single datapath anyway...
06:36:24 <yamamoto> soichi: do you have numbers?
06:37:41 <soichi> excuse me, but what do you mean "numbers"?  number of datapath?
06:37:58 <vnyyad> soichi : performance numbers i suppose
06:38:22 <yamamoto> yes, benchmark results
06:38:28 <soichi> currently, we does not have.
06:39:23 <soichi> We are under planning to mesure.
06:39:24 <yamamoto> ok.  my guess is there isn't much difference performance-wise.
06:40:17 <kaz> I thnk it must have complex flow entries.
06:41:44 <yamamoto> kaz: userland flows, i agree.  i suspect no much difference in kernel flows though.
06:43:34 <kaz> Sorry, what is "kernel flows"?
06:43:56 <yamamoto> kaz: odp flows (vs openflow flows)
06:44:32 <trinaths> Hi
06:45:14 <yamamoto> kaz: are you familiar with ovs implementation?
06:45:51 <kaz> I can't answer right now.
06:48:27 <vnyyad> anil has explained the reasons behind the current design in his mail yesterday
06:48:31 <reedip> Hello, I saw in  Media: OperationsFromGUI_20151116-01.pdf that the TAP CLIs are supported. But I couldnt find any option in the Neutron Client
06:49:50 <yamamoto> reedip: currently taas uses a dedicated "taas" command.
06:50:09 <soichi> vnyyad: yes, i am trying to understand the backgroud.
06:50:23 <vnyyad> reedip: Neutron client support for taas is proposed by soichi
06:51:27 <vnyyad> should we have a mail list to discuss these items
06:51:51 <trinaths> taas - is just port monitoring or helps in vm ha?
06:52:00 <yamamoto> vnyyad: what's wrong with openstack-dev?
06:53:16 <vnyyad> yamamoto: we can use it, i guess we add [taas] to the subject
06:53:33 <yamamoto> trinaths: just monitoring.  it might be able to be used for some kind of HA but i have no idea.
06:53:36 <yamamoto> vnyyad: sure
06:53:50 <reedip> yamamoto, vnyyad: It would be good if the following agendas are discussed on emails, so that everyone in loop knows whats happening. And I agree with yamamoto  that the Openstack devlist with [taas] can be used
06:54:23 <vnyyad> reedipt: sure, that will be done from now on
06:54:30 <dedery> shouldn't it be [neutron][taas]?
06:54:43 <yamamoto> +1 [neutron][taas]
06:54:47 <reedip> yeah, [neutron][taas] sounds better
06:54:47 <trinaths> yamamoto: okay. what more can be done using this monitoring?
06:57:34 <soichi> +1
06:57:37 <kaz> +1
06:58:03 <vnyyad> trinaths: currently it is for mirroring the traffic from a particular port to another port, what you do with the mirrored traffic is not covered under taas right now
06:59:48 <vnyyad> +1 [neutron][taas] it will be
07:00:20 <yamamoto> kaz: soichi: are you going to submit horizon patch?
07:03:18 <yamamoto> #info we'll use [neutron][taas] on openstack-dev
07:03:20 <soichi> No, it is just a first prototype. Before submit to horizon, we would like to share in TaaS project.
07:03:46 <trinaths> vnyyad: can you be more clear on "particular port to another port" *
07:04:29 <vnyyad> trinaths: from one neutron port of another neutron port (which is call the tap service port)
07:05:17 <dedery> trinaths: both ports belong to the same tenant
07:05:34 <vnyyad> soichi: thanks for the work on horizon support for TaaS
07:05:53 <yamamoto> +1
07:05:59 <vnyyad> dedery: +1
07:06:06 <irenab> dedery: this maybe the case, that admin wants to mirror tenant traffic
07:06:34 <trinaths> vnyyad dedery: confused
07:08:15 <vnyyad> irenab: In the current implementation the admin has to create a user under the tenant to mirror flow
07:08:16 <dedery> irenab: i'm aware of the use-case (especially lawful intercept) from what i read in the bp there was quite a discussion about this
07:09:01 <dedery> irenab: not sure if the discussion reached any actionable conclusion
07:09:21 <vnyyad> irenab: this was done to keep tenant in the loop that their traffic is being monitored by the admin, mainly for privacy
07:09:57 <vnyyad> dedery: yes it was discussed at length during the spec reviews
07:13:07 <vnyyad> trinaths: Please refer to https://github.com/openstack/tap-as-a-service and the spec at https://review.openstack.org/#/c/96149/8/specs/juno/tap-as-a-service.rst
07:13:54 <yamamoto> how about finishing agenda items before having random discussions?
07:14:07 <dedery> yamamoto: +1
07:14:10 <irenab> dedery: fair enough
07:14:11 <vnyyad> yes sure
07:14:13 <vnyyad> :)
07:14:14 <trinaths> vnyyad: thank you. will go through them
07:14:28 <yamamoto> vnyyad: can you use #topic to avoid sidetracking?
07:14:48 <vnyyad> yamamoto: i will
07:15:24 <vnyyad> #topic isolation of production and mirrored traffics
07:16:38 <soichi> We proposed a dedicated tunnel to carrying mirrored traffic.
07:17:32 <trinaths> soichi: can you provide the link ?
07:17:49 <yamamoto> #link https://wiki.openstack.org/w/images/7/78/TrafficIsolation_20151116-01.pdf
07:18:14 <trinaths> thank you
07:18:35 <soichi> I didn't mentioned in the document,
07:18:59 <vnyyad> soichi: this might be helpful if there is a separate physical network to carry the mirrored traffic
07:19:22 <soichi> definitely
07:19:27 <vnyyad> but if uses the same underlying physical network what might be the advantage
07:20:24 <vnyyad> The design was not to introduce more tunnels then already there
07:20:45 <soichi> Although it depends on the physical configuration of a host, we want to map trafiics to different pysical NIC
07:22:12 <vnyyad> soichi: in cases of having a different physical networks i see a good use of the proposed design, one network for production traffic and another for mirrored and admin and other traffic
07:22:33 <kaz> I may be easy to control QoS.
07:24:31 <vnyyad> any thoughts on this from others
07:25:54 <yamamoto> i wonder how much it complicates the code
07:26:27 <vnyyad> we have 5 mins in the meeting, i propose that we bring this up for discussion in the next meeting too along with discussing this on the mailing list
07:26:54 <soichi> we would like to share the source code we have implemented.
07:27:09 <yamamoto> soichi: great.
07:27:24 <yamamoto> vnyyad: +1
07:27:24 <vnyyad> soichi: that would be good
07:28:28 <yamamoto> let's move on quickly
07:28:33 <reedip> soichi: +1
07:28:37 <vnyyad> On the Dashboard support for TaaS, The effort is really appreciated and thanks for that
07:28:54 <soichi> tahnks
07:29:00 <yamamoto> #link https://wiki.openstack.org/w/images/2/2e/OperationsFromGUI_20151116-01.pdf
07:29:25 <reedip> I found only Create and Delete options, is an Update option viable?
07:29:25 <vnyyad> Anil in his mail has some good suggestions on this
07:31:07 <vnyyad> reedip there is not option for update yet in the TaaS API, we currently support create and delete for Tap Service and Tap Flow
07:31:26 <reedip> ok
07:31:33 <yamamoto> reedip: i guess it isn't too important as only updatable fields akkkre name and description
07:31:47 <vnyyad> i have to end the meeting now, but will add the items remaining to the next meeting agenda
07:31:52 <reedip> yamamoto: I was also thinking in the similar lines
07:32:18 <vnyyad> any other business
07:32:43 <yamamoto> vnyyad: sure.  and/or discuss on ML.
07:32:53 <vnyyad> ok then
07:32:58 <soichi> thank you
07:33:01 <vnyyad> Bye everybody
07:33:03 <reedip> Please share Anil's comments on the ML
07:33:08 <yamamoto> thank you
07:33:10 <kaz> bye
07:33:10 <dedery> thank you all
07:33:11 <dedery> bye
07:33:12 <reedip> that would allow everyone to come in sync
07:33:18 <vnyyad> reedip: we will
07:33:23 <vnyyad> bye
07:33:25 <reedip> +1 , tc :)
07:33:31 <vnyyad> #endmeeting