15:00:36 <ad_ri3n_> #startmeeting fog-edge-massively-distributed-clouds
15:00:37 <openstack> Meeting started Wed Jan 17 15:00:36 2018 UTC and is due to finish in 60 minutes.  The chair is ad_ri3n_. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:38 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:41 <openstack> The meeting name has been set to 'fog_edge_massively_distributed_clouds'
15:01:04 <ad_ri3n_> #chair  ad_ri3n_
15:01:04 <openstack> Current chairs: ad_ri3n_
15:01:05 <ad_ri3n_> #chair  ad_ri3n_
15:01:05 <openstack> Current chairs: ad_ri3n_
15:01:10 <ad_ri3n_> arg…
15:01:32 <ad_ri3n_> so let's see who join
15:01:35 <ad_ri3n_> #topic roll call
15:01:41 <parus> Good day!
15:01:51 <ad_ri3n_> (best wishes BTW, if I'm right this is our first meeting ;))
15:01:53 <ad_ri3n_> Hi Parus
15:02:04 <kgiusti> o/
15:02:28 <ansmith> o/
15:02:41 <ad_ri3n_> #link https://etherpad.openstack.org/p/massivfbely_distributed_ircmeetings_2018 line 29 agenda
15:03:01 <ad_ri3n_> sorry wrong link
15:03:16 <ad_ri3n_> #link https://etherpad.openstack.org/p/massively_distributed_ircmeetings_2018 line 29 agenda
15:03:19 <avankemp> o/
15:03:38 <ad_ri3n_> Please go through the announcement section
15:03:43 <ad_ri3n_> If you have question please feel free
15:03:44 <ad_ri3n_> ;)
15:04:04 <msimonin> o/
15:04:09 <alisanhaji> hi
15:04:11 <ad_ri3n_> new resolutions for 2018, let's try to be more fruitful
15:04:12 <ad_ri3n_> ;)
15:04:15 <ad_ri3n_> Hi alisanhaji
15:04:20 <msimonin> hey alisanhaji :)
15:05:08 <ad_ri3n_> ok so last minute for asking questions regarding the news :)
15:05:28 <alisanhaji> What are the news?
15:05:42 <ad_ri3n_> sorry alisanhaji
15:05:51 <ad_ri3n_> please give a look to our etherpad:
15:05:55 <ad_ri3n_> https://etherpad.openstack.org/p/massively_distributed_ircmeetings_2018 line 29
15:06:05 <ad_ri3n_> and BTW please add your name in the attendees section
15:06:06 <ad_ri3n_> thanks
15:06:30 <ad_ri3n_> maybe I can just add that we have been working with rcherrueau the whole day trying to achieve a gap analysis
15:06:48 <ad_ri3n_> of existing code base w.r.t. edge requirements
15:07:05 <rcherrueau> o/
15:07:12 <msimonin> hi rcherrueau
15:07:16 <parus> Is there an etherpad?
15:07:17 <rcherrueau> hi msimonin
15:07:22 <msimonin> parus:  +1
15:07:24 <alisanhaji> hi rcherrueau
15:07:27 <ad_ri3n_> I'm not sharing it yet as it is really crappy from the formatting viewpoint now
15:08:26 <rcherrueau> hi alisanhaji
15:08:43 <parus> If we are still in news. I can point out two new community initiatives.
15:09:02 <ad_ri3n_> sure please go ahead
15:09:06 <ad_ri3n_> (and add it to the pad ;))
15:09:09 <parus> 1- The ONAP usecase team has started a group looking at Edge computing.
15:09:16 <ad_ri3n_> great
15:09:20 <ad_ri3n_> some links?
15:09:55 <parus> 2- The IEEE, withing 5g.ieee.org has started a group called EAP (Edge Automation Platform) that is developping a framework for Edge.
15:10:20 <ad_ri3n_> thanks
15:10:20 <parus> I will dig out the links and post on etherpad.
15:10:23 <ad_ri3n_> good to not
15:10:26 <ad_ri3n_> to know
15:10:36 <ad_ri3n_> I'm wondering how we can combine our effort
15:10:56 <ad_ri3n_> (we should really fine a way to do not reinvent the wheel again and again)
15:11:34 <ad_ri3n_> ok nothing else?
15:11:48 <ad_ri3n_> So let's switch to the first topic:
15:11:58 <ad_ri3n_> #topic ongoing-action-AMQP
15:12:06 <ad_ri3n_> guys, the floor is yours
15:12:18 <msimonin> so
15:12:45 <kgiusti> dragonflow - I've been sidetracked on other stuff but will begin eval work this week.
15:13:07 <msimonin> great
15:13:34 <kgiusti> otherwise that's about all I have this week (aside from a fix or two for the ombt2 tool)
15:14:03 <ad_ri3n_> thanks kgiusti
15:14:05 <ad_ri3n_> msimonin:  ?
15:14:12 <ad_ri3n_> do you want to add something particular?
15:14:18 <msimonin> From the massively distributed RPCs test plan side
15:15:01 <msimonin> Until now we are iterating on several tools : ombt2, the test framework, plugins
15:15:09 <msimonin> plugins for collectd*
15:15:25 <msimonin> I think we diverge a bit from the innitial goal
15:15:48 <msimonin> which was to implement the test case (https://docs.openstack.org/performance-docs/latest/test_plans/massively_distribute_rpc/plan.html)
15:16:36 <ad_ri3n_> ok
15:16:37 <msimonin> We'll need to refocus a bit during the next few weeks
15:16:58 <msimonin> especially get back to the initial massively distributed aspects of the test case
15:17:02 <ad_ri3n_> ok no problem. I guess if did this choice because it was/is relevant?
15:17:19 <ad_ri3n_> s/if did/ if you did
15:17:40 <msimonin> Yes, all the tools are better than before, the framework is strong
15:18:04 <ad_ri3n_> cool
15:18:17 <ad_ri3n_> You need to provide us a short overview of this stack ;)
15:18:21 <ad_ri3n_> a white paper :-P
15:18:43 <ad_ri3n_> anything else?
15:18:55 <msimonin> #link https://github.com/msimonin/ombt-orchestrator
15:19:04 <msimonin> this a good entry point I guess
15:19:35 <ad_ri3n_> (@msimonin it seems someone forgot to put the link toward the first results regarding rabbit)
15:19:43 <ad_ri3n_> line 54 in the pad
15:20:21 <ad_ri3n_> thanks
15:20:21 <msimonin> ad_ri3n_:  done
15:21:12 <ad_ri3n_> ok in a few words
15:21:23 <ad_ri3n_> what are the first lessons you discovered?
15:21:52 <msimonin> We made all the experiments using rabbitmq (so centralized message bus)
15:22:39 <ad_ri3n_> and?
15:22:40 <msimonin> This are more baseline study for now
15:22:42 <ad_ri3n_> ok
15:23:25 <ad_ri3n_> ok so more infos next time I guess?
15:23:58 <msimonin> You are greedy :)
15:24:03 <ad_ri3n_> :-P
15:24:17 <ad_ri3n_> ok so let's move to the next topic
15:24:33 <ad_ri3n_> #topic ongoing-actions-cockroach
15:24:41 <ad_ri3n_> @rcherrueau @Marie_
15:24:44 <ad_ri3n_> ?
15:25:19 <rcherrueau> Nothing really new regarding nova. WIP ... (still struggling with the correlated subqueries)
15:25:19 <rcherrueau> 
15:26:01 <rcherrueau> Good news btw, Marie_ (o/ Marie_ ?) started an internship at Inria and she gonna work on cockroachdb perf
15:26:31 <rcherrueau> We plan to do perf analysis of Keystone + Galera vs. Keystone + CockroachDB
15:27:00 <Marie_> Hi everyone :)
15:27:04 <rcherrueau> (maybe we will reuse the framework of msimonin & co ;)
15:27:11 <msimonin> I  can help writing the experiments :)
15:27:37 <rcherrueau> We plan to redo scenario defined in the performance working group
15:27:42 <avankemp> so can I :) (Hi Marie !)
15:28:11 <rcherrueau> But if you have a specific scenario in mind, and want us to test is, do not hesitate
15:28:21 <rcherrueau> to test it*
15:28:28 <parus> @rcherrueau: is there anything we can learn from the paper that Allison shared
15:28:28 <ad_ri3n_> Thanks, maybe one can mention also that you push a blog related to the edge activities within the discovery and the first ticket is around the implementation cockroach-based keystone
15:28:36 <parus> #link http://cnp.neclab.eu/projects/lightvm/lightvm.pdf
15:29:05 <ad_ri3n_> yes parus ?
15:29:10 <parus> In particular the part where they remove database?
15:29:23 <ad_ri3n_> @rcherrueau can you please c/p the link to the post
15:29:25 <msimonin> Should we go discuss at the PTG with Keystone teamp ?
15:29:42 <rcherrueau> ad_ri3n_: yep, discovery blog
15:29:56 <rcherrueau> #link https://beyondtheclouds.github.io/blog/
15:30:38 <rcherrueau> parus: I have to take a look at the paper, thanks for the link
15:31:00 <msimonin> rcherrueau:  We could ping Keystone folks to gather use cases
15:31:59 * ad_ri3n_ is looking for the section dealing with DB stuff (any hint welcome ;))
15:32:19 <rcherrueau> msimonin: you are right, lemme add this as a TODO on my side.
15:32:43 <ad_ri3n_> parus: I cannot see this paragraph
15:32:47 <ad_ri3n_> can you point me?
15:33:08 <parus> Give me a minute! I am not as fast as you are :-)
15:33:28 <parus> section 5.1 No XenStore
15:33:44 <ad_ri3n_> thanks
15:33:47 <ad_ri3n_> I will give it a look
15:33:49 <ad_ri3n_> thanks
15:34:35 <ad_ri3n_> I'm not sure (I may be wrong as I just went through the paper quickly)
15:34:53 <ad_ri3n_> but the paper looks to be focused on a single hypervisor
15:35:14 <ad_ri3n_> I will read it deeply and let's see what we can learn from it.
15:35:22 <parus> It is! but Allison's point is that we can learn from their methodology.
15:36:02 <parus> Their results are irrelevant to Openstack. The methodology can be reused.
15:36:15 <ad_ri3n_> ok thanks for claryfing and in any case, reading SOSP papers is never waisted time ;)
15:36:23 <ad_ri3n_> ok anything else regarding the cockroach stuff
15:36:23 <ad_ri3n_> ?
15:36:29 <rcherrueau> nop
15:36:44 <ad_ri3n_> #ongoing-action-usecases
15:36:57 <ad_ri3n_> parus:  do you want to mention a few words?
15:37:13 <parus> yes.
15:37:32 <parus> Things are moving with the help of the foundation on use cases.
15:37:53 <parus> I would like to get this team lined up behind this effort.
15:37:53 <ad_ri3n_> for sure, there are discussions
15:38:27 <parus> For instance, the etherpad above gives details on a use case for a medium size scenario..
15:38:29 <ad_ri3n_> what we can work on is to try to identify expected features in terms of functions
15:38:46 <parus> I would like to use that same model and flesh out the other use cases.
15:38:58 <ad_ri3n_> and conduct a gap analysis w.r.t the Openstack code base (and the different deployment scenarios)
15:38:59 <ad_ri3n_> ?
15:39:03 <ad_ri3n_> what do you think parus?
15:39:19 <parus> Yes
15:39:29 <ad_ri3n_> we select one use-case (let's say medium) and dive into details
15:39:58 <ad_ri3n_> a few lines from the work we did today with rcherrueau
15:40:14 <parus> yes?
15:40:22 <ad_ri3n_> #link https://etherpad.openstack.org/p/massively_distributed_ircmeetings_2018 line 69
15:41:47 <parus> Who would like to help?
15:42:31 <ad_ri3n_> so from our side we hope to have a first proposal to be discussed by the end of the week
15:43:22 <ad_ri3n_> parus:  regarding your comment in the pad
15:43:40 <ad_ri3n_> we think rather in terms of expected features from the naive one to most advanced
15:43:51 <parus> OK
15:44:04 <ad_ri3n_> openstack footpring for instance is something we should consider but we believe after
15:44:21 <ad_ri3n_> Level 1:
15:44:30 <ad_ri3n_> I want to be able to perform normal cloud operations on one site
15:44:49 <ad_ri3n_> Level 2: I want to be able to perform nomral/usual cloud operation on another site
15:44:59 <ad_ri3n_> Level 3: network split brain comes into the picture
15:45:15 <ad_ri3n_> Level 4: mesh applications (cross network vpn, relocation of workloads)
15:45:24 <ad_ri3n_> Level 5: can be space limitations
15:45:34 <ad_ri3n_> what do you think?
15:45:36 <parus> What about storage?
15:45:46 <ad_ri3n_> you mean SWIFT like solution?
15:45:57 <ad_ri3n_> or storage for the workloads (i.e. to be able to boot them)?
15:46:05 <parus> I see level 4e has something.
15:46:10 <ad_ri3n_> Level 1.f.
15:46:11 <ad_ri3n_> too
15:46:29 <ad_ri3n_> level 1 normal cloud, I want to use remote attached volume in one edge site
15:47:08 <parus> and 4c would cover chaining?
15:47:17 <ad_ri3n_> I believe that even before space limitations (i.e. openstack footprint
15:47:23 <ad_ri3n_> yes
15:47:25 <ad_ri3n_> exactly
15:47:36 <ad_ri3n_> Level 5 can be related to all advanced admin operations (zero provisionning
15:47:37 <ad_ri3n_> etc…)
15:48:06 <parus> How about failure/recovery?
15:48:21 <ad_ri3n_> failure is cover in Level 3
15:48:21 <parus> 1000 nodes coming back online at the same time.
15:48:24 <ad_ri3n_> recovery is not cover yet
15:48:33 <ad_ri3n_> :D
15:48:36 <ad_ri3n_> advanced operations
15:48:38 <ad_ri3n_> Level 5
15:48:38 <ad_ri3n_> :D
15:48:46 <ad_ri3n_> you are the FEMDC killer ;)
15:48:53 <parus> That sounds like a plan.
15:49:01 <ad_ri3n_> and then we conduct a gap analysis
15:49:19 <ad_ri3n_> according to
15:49:22 <ad_ri3n_> vanilla openstack
15:49:28 <ad_ri3n_> with remote compute nodes
15:49:39 <ad_ri3n_> federated keystone
15:49:41 <ad_ri3n_> regions
15:49:43 <ad_ri3n_> cells
15:49:57 <parus> Great plan!
15:50:00 <ad_ri3n_> I hope we can present a clean etherpad soon
15:50:04 <ad_ri3n_> If you want to help us ;)
15:50:09 <ad_ri3n_> you are more than welcome
15:50:19 <ad_ri3n_> (and if you have a better idea, same ;))
15:50:26 <parus> Let me get on a plane :-)
15:50:37 <ad_ri3n_> :-)
15:50:54 <ad_ri3n_> ok I will keep you in touch soon
15:51:05 <parus> +1
15:51:06 <ad_ri3n_> ok let's move forward I would like to discuss a bit the other points
15:51:36 <ad_ri3n_> is there any dragonflow folks?
15:51:41 <ad_ri3n_> I'm not sure
15:52:07 <ad_ri3n_> ok so let's move to the open discussion (I have nothing to add regarding the other points)
15:52:24 <ad_ri3n_> #topic ongoing-action-dragonflow
15:52:26 <ad_ri3n_> Not discuss
15:53:01 <ad_ri3n_> #topic ongoing-acton-p2p-ericsson
15:53:06 <ad_ri3n_> not discuss
15:53:12 <ad_ri3n_> #topic open discussion
15:53:15 <msimonin> (dragonflow )I think kgiusti filled the pad l53
15:53:36 <ad_ri3n_> I want to higlight the vancouver CFP once again (already highlighted in the news)
15:53:47 <ad_ri3n_> it would be great if we can identify
15:53:59 <ad_ri3n_> what presentations can be submitted by the FEMDC group
15:54:11 <ad_ri3n_> can we make a first roundtable ?
15:54:25 <ad_ri3n_> @rcherrueau cockroach?
15:54:48 <ad_ri3n_> @msimonin @kgiusti et al. AMQP ?
15:54:51 <rcherrueau> ad_ri3n_: yes it could be a good idea.
15:55:01 <ad_ri3n_> @parus use-case/gap analysis?
15:55:05 <ad_ri3n_> anything else?
15:55:51 <parus> Anything on performance?
15:55:59 <ad_ri3n_> the AMQP evaluations I gues
15:56:04 <kgiusti> msimonin: perhaps something around the AMQP eval?
15:56:07 <ad_ri3n_> cockroach can also incluse some performance
15:56:14 <ad_ri3n_> please see the pad
15:56:17 <ad_ri3n_> I put some proposals
15:56:21 <rcherrueau> parus: same for cockroach I guess
15:56:40 <msimonin> kgiusti:  oh yes this would be great
15:56:40 <ad_ri3n_> kgiusti: is there a link toward the PTG schedule please ?
15:56:41 <kgiusti> I'd be willing to help present (if I can get travel)
15:57:07 <msimonin> kgiusti:  we can discuss this during our weekly session
15:57:20 <kgiusti> ad_ri3n_: I'm not aware of a detailed schedule yet - ansmith?
15:57:23 <ad_ri3n_> @parus will you come to the PTG?
15:57:33 <parus> I am not confirmed yet. Trying to !
15:57:44 <ansmith> I have not seen
15:58:08 <ad_ri3n_> ok
15:58:10 <ad_ri3n_> one minute left
15:58:15 <ad_ri3n_> anything else to mention?
15:58:18 <kgiusti> ad_ri3n_: but just in terms of my availability I need to help run the oslo meeting (which I think is during the first two days)?
15:58:18 <ad_ri3n_> (please put it in the pad)
15:58:28 <ad_ri3n_> kgiusti:  shame :(
15:58:34 <ad_ri3n_> it is a pitty we do not have the schedule
15:58:41 <msimonin> ad_ri3n_: ttx sends a mail on the ml regarding the schedule (still struggling with it I gues)
15:58:46 <ad_ri3n_> because it is really difficult to synchronise the effort
15:58:53 <ad_ri3n_> msimonin:  thanks
15:58:57 <ad_ri3n_> ok time to conclude
15:58:59 <parus> thanks
15:59:22 <ad_ri3n_> ok thakns
15:59:26 <ad_ri3n_> talk to you next time
15:59:27 <ad_ri3n_> or by mail
15:59:29 <ad_ri3n_> #endmeeting