05:33:31 #startmeeting taas 05:33:32 Meeting started Wed Sep 21 05:33:31 2016 UTC and is due to finish in 60 minutes. The chair is soichi. Information about MeetBot at http://wiki.debian.org/MeetBot. 05:33:33 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 05:33:35 The meeting name has been set to 'taas' 05:33:45 Hi 05:33:50 #topic: Inclusion in Governance and Neutron Stadium 05:33:53 hi vnyyad 05:34:04 armax : ping ? 05:34:32 I think he is busy ,but it would be great if we can talk to him about this. 05:34:33 could not ping him yesterday, will try today 05:34:46 actually he is on now 05:35:42 the basic query is , if we should apply for governance first, and then move onto neutron stadium? 05:35:54 yes 05:36:19 can we ask this in the neutron mailing list and try to get some opinion or answer? 05:37:04 vnyyad : I think we can . 05:37:48 reedip: ok... shall we try that then 05:38:20 anil: any thoughts 05:38:31 vnyyad : I think we can 05:39:25 reedip: so let me put in a mail today 05:40:33 vnyyad , lets see if anil_rao has some inputs for this? 05:40:39 +1 05:41:19 hi 05:41:30 yamamoto: hi 05:43:12 anil_rao: do you have any thoughts? 05:45:16 let's go on next topic 05:45:27 #topic: Performance measurement (progress report) 05:45:49 We uploaded a new document about the performance measurement. 05:46:09 #link: https://wiki.openstack.org/w/images/e/e9/Increasing_Send_Packets20160921.png 05:46:28 Slide 1 shows a performance of the original (community) implementation. 05:47:14 The horizonal axis means # of send packets by "iperf -b Npps" 05:47:49 the vertical axis means # of received packets and mirror packets. 05:48:22 It seems that overhead of mirroring is small if the number of send packets is less than about 70k pps. 05:49:03 Slide 2 shows of the flow based tunneling version we implemented. 05:49:03 packet size: 64byte (lleft graph) and 128byte (right graph) 05:50:13 by flow based you mean a filtering based on 5 tuple... before mirroring ? 05:52:16 we did not use filtering 05:53:17 ok can you just elaborate on what flow based meant here 05:53:54 We implemented flow based tunneling for separation the production traffic and mirror traffic. 05:54:57 ha ok... thanks for the clarification 05:56:58 Slide 3 shows a comparison of when packet length is 1400Byte. 05:57:42 the left graph is original version, and the right is flow based tunneling version. 05:58:46 It seems that the received packets with mirror and mirror packets are almost half of the packets without mirror(blue line) 05:59:13 It is because the physical network was saturated. 06:00:54 Each data of the right graph is almost same bacause of the separation between the production and the mirror traffic 06:01:03 to different physical network. 06:02:36 so everything seem working as expected? 06:03:55 yamamoto: yes 06:04:06 as a result we can say that trafic separation between production and mirror is effective 06:04:43 +1 06:06:13 kaz and soichi: thanks guys! 06:07:05 thank you, we will make additional measurements 06:07:48 can we go on next topic? 06:08:34 #topic: Open Discussion 06:08:55 Good news: TaaS presentation proposal was selected for the vBrownBag Tech Talk in Barcelona Summit 06:08:57 on 27th Oct., Wedensday, 15:15-15:30 06:10:37 we will present performance characteristics of TaaS 06:10:39 Wed is 26 06:10:58 yamamoto: yes, 26th 06:12:39 #link: http://vbrownbag.com/2016/09/vbrownbag-techtalks-at-openstack-barcelona/ 06:12:54 just FYI 06:15:15 do you have any other topics or update for today? 06:17:19 nothing from me 06:17:31 nothing here too 06:17:34 i have no more topics. 06:18:46 sorry, I was out 06:19:13 reedip: do you have any topic or update? 06:19:50 soichi : nothing much as such , but I wanted to know why we cannot create BPs for TaaS 06:20:04 vnyyad : Launchpad doesnt allow us to create Blue prints currently 06:20:11 for Tap-as-a-Service 06:20:38 reedip: let me take a look and geet baack to you 06:20:52 vnyyad: +1 06:20:56 okkkk , vnyyad : I wiiil waiit :D 06:21:34 :) 06:22:08 anything else? 06:23:26 it seems we can close today's meeting 06:23:31 thank you guys, see you next week 06:23:47 #endmeeting