 Hi there, this is Heidi George-Rathaway with the OpenStack Foundation. I'm here with the Metaca Design series featuring John Garbet from NOVA. John, tell us a little bit about yourself. Thank you for having me. So I'm John Garbet. I'm currently a Principal Engineer at Rackspace working on the public cloud there. I was NOVA-PTL for the last cycle for the Liberty Ticle and also NOVA-PTL from Metaca. You're also known as the Tuba Guy. I am. My RSE next, John the Tuba Guy. Let's talk a little bit about the hot topics that you discovered in Tokyo at the Design Summit. What were some of the decisions and outcomes from these discussions? So one of the things we spent a lot of time on was looking at the resource modeling. So how NOVA and the scheduler interacts and sort of coming up with a way in which we can model that interaction and get a good versioned API there. There were lots of other good conversations sort of focusing on specific issues that people have come up with. We had some on-conference sessions looking at drilling into the specific issues, see how we get forward on that. That was really productive I think. One of the other things I guess is stuff that wasn't a hot topic. There was a lot of stuff that we agreed ahead of time using the spec process and the spec-less blueprint process and got it all locked into place before the summit so we could really focus on things that needed the in-person discussion and get down to the details on those so we can get everyone moving forward as fast as possible. So tell us more about some of the planning and particularly the user needs that you had identified as the problems you're trying to solve. Yeah, let me sort of give you a quick whistle-stop tour of some of the big highlights of things that we spoke about. One was config options. We have an amazing number of config options, at least 800, almost uncountable. We need to get those down. We need to work out what they all do and get that documented. We had a developer go digging and he sort of realized how bad it was in some of these areas so we need to go down that route and try to sort that out. There's a whole host of other little things. So for example, we have a notification system to sort of integrate with internal systems. We found a way of versioning that and trying to evolve that so we can get that to a better state and get some documenting going. There's some great work to try and adopt OS Info. There's a new package that can help with getting better default configuration values for images to make the image metadata setting easier. There's some interesting NFV use cases that we were discussing. There's some new physical function pass-through for SROV. There's hierarchical quotas came up. We think we found a way forward to get that in. Looking at manila integration, most folks want to isolate their connections from the network. A whole heap of really interesting things. I guess there's a lot more sort of things involving cross-projects and starting to make progress between the multiple projects. One highlight from that kind of set of stuff was working with people from Trove and other kind of projects where they need to use NOVA so they need to protect the server that they have running with the NOVA. So we're looking at ways forward on moving forward with that. Tell us more then about the priorities that came out of the Design Summit. What do you see as maybe the top three or four priorities for your new features and enhancements going forward? So one thing we started doing NOVA over the past few releases is actually as a community using one or two design summit sessions, getting together and sitting down and agreeing what we want to focus on and setting up the process to make sure that we can actually focus on that. Naturally, you asked for three or four and being a developer, I came up with five just to annoy you. So the ones that we have right now at first is Live Migrate. There's no particular order, but Live Migrate is one of those operations that have been a problem point, largely due to a lack of upstream testing. So we're working on doubling down on that. There's a whole group of people from all sorts of different companies getting together and focusing on that, which is great. So we want to support that to make it a priority. There's quite a few things that we've carried over from the last release. So these are multi-cycle things that are just taking time. One is the API work, so we're going to focus on API documentation and making sure that the service catalog, well, making sure that the API works well for the service catalog. There's another piece, which is Cells. So Cells v2 work. So that, again, is going to be a focus for this cycle. A piece that there was a lot of interest on was OSVIF Lib, that's another priority, which is basically trying to make sure that we have a solid interface between Nova and Neutron to do with all the VIF plug-in. So we're going to work on that. And that's really important to support Neutron's big stadium of projects and make sure that we keep the velocity on the VIF plug-ins. Another piece, which I was talking about earlier, is the scheduler. So we're going to keep working on the scheduler and getting that interface solid. So yeah, it's live migrate, the API, Cells, OSVIF Lib, and the scheduler. Well, one of the ways that we try to connect the dots among projects is by defining project themes. And so I wondered if there were a few key themes or maybe one theme that you see that your project will really achieve in Mitaka. That's a good question. One of the things I've done is I've sat down and looked at the themes of the product working group are using, the scale of relative resiliency and all these kind of things. And honestly, Nova has little pieces that actually hit on each one of these. So I'll dig into a few of these just to add some color to that list of priorities. So scalability. One of the things that Cells is actually helping with is trying to get to that next level of scale. We need to make sure a single Cells scale as well that also have ways of attaching these scales. Expanding deployments. So it also adds to the manageability pieces while at the same time. Because you can deploy individual Cells, test them, get them ready, and then add them into production. So you're sort of deploying in these little units that you keep adding. The Cells are super useful for that. Another piece is probably resiliency. The live migrate work, that's really kind of digging into there. There's lots of other things, sort of resiliency, technical debt, kind of pieces that are coming up. The live migrate, but as well as keeping the API is during upgrade, keeping the API up during upgrade, tackling things like quotas. As I mentioned, the hierarchical quotas. There's lots of pieces there. Modularity is a big thing. The RSVIF lib, making sure that the Nova Neutron interface is good. Making sure that the schedule interface is good so we can start adding more support in there. Lots of work on those fronts. And again, lots of work on the API and making sure it's documented so people can use that on the interoperative front. So, yeah, lots of things. John, thank you so much for taking time with us. We really appreciate it. John, the two will die. Thank you very much. See you later.