16:03:37 <sgordon_> #startmeeting nfv
16:03:38 <openstack> Meeting started Thu Oct 16 16:03:37 2014 UTC and is due to finish in 60 minutes.  The chair is sgordon_. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:03:38 <cloudon> hi
16:03:40 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:03:42 <openstack> The meeting name has been set to 'nfv'
16:03:43 <sgordon_> #topic roll call
16:03:45 * sgordon_ here
16:03:54 <cloudon> +1
16:04:00 * beagles here
16:04:13 <sgordon_> i suspect this will be a relatively quick affair
16:04:18 <sgordon_> #link https://etherpad.openstack.org/p/nfv-meeting-agenda
16:04:27 <sgordon_> #topic Review actions
16:04:46 <sgordon_> #info imendel was to touch base with Michael (ETSI liaison) and clarify resource catalog requirement
16:04:56 <sgordon_> no sign of imendel today so i will carry that over
16:05:08 <sgordon_> #action imendel to touch base with Michael (ETSI liaison) and clarify resource catalog requirement
16:05:29 <sgordon_> #info steveg was to get BoF space
16:05:30 <sgordon_> ok
16:05:40 <sgordon_> so there was some discussion wrt BoFs on the community list
16:05:58 <sgordon_> and the sum total is there is still not going to be any formal way to reserve/book space other than the pods
16:06:09 <sgordon_> so we will have to do something similar to atlanta
16:06:25 <sgordon_> albeit attempting to lock in at least a time a little earlier...
16:06:41 <sgordon_> any strong preferences from those here?
16:06:59 <sgordon_> #action sgordon to email list regarding NFV BoF and preferred dates/times during summit
16:07:16 <sgordon_> #info May need to determine "place" for BoF on the ground based on POD availability etc.
16:07:24 <cloudon> no particular preference
16:08:03 <sgordon_> me either, just the knowledge that it is going to be a challenge to cater to everyone
16:08:07 <sgordon_> :)
16:08:13 <sgordon_> #topic juno status
16:08:19 <sgordon_> #info 'it happened'
16:08:32 <sgordon_> i believe thierry finalized the release from a technical perspective today
16:08:42 <sgordon_> so right on schedule
16:08:53 <sgordon_> #topic kilo status
16:09:12 <sgordon_> #info specs repositories (and code repositories for that matter) are well and truly open for business
16:09:23 <sgordon_> #info https://blueprints.launchpad.net/nova/+spec/input-output-based-numa-scheduling has been resubmitted
16:10:11 <sgordon_> #info Summit planning etherpads are shaping up here: https://wiki.openstack.org/wiki/Summit/Planning
16:10:19 <cloudon> is luke gorrie here to comment on his Snabbswitch accel data plane & Kilo?
16:10:30 <sgordon_> i cant see him
16:10:35 <sgordon_> his usual nick is lukego i think
16:11:01 <sgordon_> #info Will Luke Gorrie be resubmitting Snabbswitch accelerated data plane work for Kilo?
16:11:07 <sgordon_> #action sgordon to follow up with luke
16:11:12 <cloudon> ta
16:11:42 <sgordon_> #topic other discussion
16:11:58 <sgordon_> so that's my story, anything else anyone wants to cover today?
16:12:05 <sgordon_> in particular any neutron specs resubmitted?
16:12:10 <marun> sgordon_: regarding snabb, we're considering a split out of vendor/plugin code in kilo
16:12:14 <sgordon_> i have mainly been trawling through nova-specs
16:12:32 <sgordon_> marun, ack
16:12:33 <marun> sgordon_: I'm not sure what advantage there would be in that case to being officially in tree vs carried outside.
16:12:51 <sgordon_> #info Split out of neutron vendor/plugin code being considered for Kilo, would impact Snabb changes
16:13:18 <sgordon_> marun, iirc his proposal got split such that there is some work on the nova side as well
16:13:20 <marun> sgordon_: I've discussed pushing at least 3 use cases this cycle
16:13:24 <sgordon_> not sure how that plays out
16:13:43 <marun> sgordon_: vlan trunking, unaddressed interfaces and port mirroring should be feasible
16:14:08 <sgordon_> #info marun discussing feasibility of vlan trunking, unaddressed interfaces and port mirroring  for kilo
16:14:14 <sgordon_> marun, thanks muchly
16:14:32 <jchapman> FYI: I/O (PCIe) based NUMA scheduling spec approved, thanks to core devs, Russell, Stephen and John
16:14:34 <marun> sgordon_: I've confirmed that the implementation footprint should be minimal enough to justify implementation despite the declared focus on technical debt.
16:14:43 <marun> sgordon_: (of those specific features)
16:14:48 <sgordon_> marun, are those likely to require summit sessions to cover or discussed via M/L and spec review
16:15:24 <sgordon_> trawling through the etherpad i think they were all added at various points but it doesnt look like the neutron one has been culled yet
16:15:29 <marun> sgordon_: definitely won't be discussed at summit, spec review and M/L should be sufficient.
16:15:34 <sgordon_> ack
16:16:16 <marun> sgordon_: they aren't terribly contentious or complex proposals, we'll just need to make sure that we're responsive to community feedback.
16:16:16 <sgordon_> #info implementation implications to be discussed via spec review and M/L (no summit session req.)
16:16:37 <russellb> jchapman: :)
16:16:37 <marun> sgordon_: Also, pushing very early in the cycle will be key to getting them in.
16:16:57 <sgordon_> #info jchapman notes that I/O based NUMA scheduling spec has actually already been re-approved
16:17:20 <sgordon_> marun, ack - i think that is a goal for everything tbh
16:17:36 <sgordon_> on the nova side how we go about actually achieving that is a pretty active discussion...
16:17:57 <sgordon_> ('that' being having a better focus throughout the release cycle rather than only when we get to M-3)
16:18:30 <marun> sgordon_: I think having engaged reviewers to ensure quick turnaround is key.
16:19:08 <marun> sgordon_: having myself and armax on board to nfv efforts in neutron should make that possible this cycle.
16:19:18 <marun> to nfv -> to champion nfv
16:21:01 <sgordon_> #info marun and armax aiming to champion nfv in neutron
16:21:23 <sgordon_> #info aiming to work with reviewers to ensure a good feedback loop exists early
16:21:28 <sgordon_> marun, thanks again
16:21:31 <armax> marun, sgordon_: we have to scope what this means though
16:21:50 <sgordon_> well from a practical standpoint
16:21:58 <sgordon_> there is nfv and there is nfv, and that's part of the problem
16:21:58 <armax> marun, sgordon_: but we should be able to tackle the low-hanging fruits for sure
16:22:08 <sgordon_> championing nfv doesnt mean championing all of the things necessarily
16:23:00 <sgordon_> but what i have aimed to do with the broader group is focus on highlighting some low hanging fruit to prioritize
16:23:35 <sgordon_> vlan trunking into vms and unaddressed interfaces/ports just keep coming up :)
16:24:39 <marun> sgordon_: Do you have visibility on features required to enable service vm orchestration by something other than neutron?
16:24:46 <marun> sgordon_: e.g. heat?
16:24:49 <sgordon_> #info need to scope what 'championing nfv' actually means, not necessarily all of the things
16:24:54 <sgordon_> marun, no i dont believe i do
16:25:12 <sgordon_> marun, there is awareness that nfv reqs are going to stretch to heat (even ignoring that case)
16:25:23 <sgordon_> but not much more imo
16:25:32 <marun> sgordon_: If the goal is to enable out-of-the-box support for service vm integration, we may need to do some prototyping.
16:25:44 <marun> sgordon_: though I was under the impression that many interested parties had already done so.
16:25:44 <sgordon_> marun, i think that's fair
16:26:00 <sgordon_> they may well have and i am just not aware of the results
16:26:35 <marun> sgordon_: I know there was an etherpad from the session in atlanta that documented the gaps.  I'll try to dig it up.
16:27:16 <marun> sgordon_: I think teams from cisco and maybe intel were hacking neutron to get what they needed and documented what they needed to do to enable their specific use-cases.
16:27:45 <marun> https://etherpad.openstack.org/p/servicevm
16:27:53 <sgordon_> #link https://etherpad.openstack.org/p/servicevm
16:28:06 <sgordon_> #info prototyping of service vm integration
16:28:22 <sgordon_> i noticed this was on the list of potential topics for summit again
16:28:59 <sgordon_> not surprisingly there is some overlap in that list with individual features we have been tracking here
16:29:11 <marun> sgordon_: I'll revisit the requirements documented in that etherpad and see if I can extract the relevant use cases.
16:29:36 <marun> sgordon_: I don't think the issue is likely to get summit session time.
16:29:41 <sgordon_> #action marun to revisit requirements from servicevm etherpad and attempt to extract relevant use cases
16:29:43 <sgordon_> right
16:30:08 <marun> sgordon_: But that doesn't preclude having an informal session on friday with interested parties.
16:30:43 <marun> sgordon_: I'm pushing for part or all of one of the neutron sessions to be devoted to lightning talks.
16:31:27 <marun> sgordon_: The idea being that topics that don't get formal session time could at least present the elevator pitch to interested parties, with the goal of collecting those people for informal meetings on friday.
16:31:43 <sgordon_> makes sense
16:32:56 <sgordon_> thanks for your input marun
16:33:13 <sgordon_> i didnt have any other topics for the agenda today
16:33:32 <sgordon_> so unless someone has a particular burning issue they would like to raise i will close the meeting
16:33:40 <sgordon_> i have a few AIs to follow up with on the M/L
16:34:19 <marun> sgordon_: thank you!
16:34:27 <sgordon_> #endmeeting nfv