 Hi, this is Pete Chadwick a member of the product working group for the open stack organization community Today, I'm here with Amrith Kumar from to Zora who is the ptl for the trove project Amrith why don't you? Introduce yourself and give us a little background on what you've been doing with open stack Sure. Thank you very much Pete My name is Amrith. I'm the ptl for the trove project in the Newton cycle I'm also the chief technology officer for tesora and we are the trove company we're a company which focuses primarily on trove development and a trove based product and we have a team that is In Cambridge, Massachusetts and Mississauga, which is near Toronto and Canada great Tell me a little bit more about what trove is and explain to people exactly what you know What the project does and what's what its overall goals are? Sure. So so trove is the open stack database in the service project. It was incubated before before the ice house release it was integrated in ice house and It basically gives users the ability to easily consume databases in an open stack cloud So if you think about it and gone are the days when you had, you know, one size fits all or a corporate database standard, if you will app developers architects They tend to use the best database for the job and that means that most enterprises now have Applications that use a variety of databases some relational some non-relational no escuel database and so on Trove is a service that provides a common set of API's with which to provision and manage the life cycle of a variety of different database technologies With this a cloud service provider, for example, or an IT organization can offer an easy-to-consume self-service database portal and users can go to that portal and provision the database of their choice and Trove will provision the database for them It provides a simple mechanism to also automate the majority of the life cycle maintenance operations for a database and A database user can now orchestrate complicated database technologies like Reputation and clustering and so on and so forth effectively to ensure data security data integrity and Easy configuration management so at a very high level enterprises can improve their stability security of databases While reducing total cost of ownership and risk. That's basically what showed us for you Okay So we just finished up the the Austin planning sessions of design summit Well, I guess it sounds almost three weeks ago What were the hot topics that? The Trove team discussed and kind of what decisions did you did you try to drive in the discussions? Okay, so we had a lot of very interesting conversations at the summit Give you a quick summary of some of these conversations There was a lot of conversations about how we should handle upgrades in Trove we had several different approaches which we investigated we discussed the spec which has been put up for review we had we had a fairly interesting conversation or set of conversations around containers and How people can orchestrate databases within containers this discussion started before the summit summit with a spec which had been proposed we continued at the summit at At the summit we had several points of views expressed one point a few was that Trove should support Alternate compute back in so at this point Trove supports Nova is a single compute back in There was a point of view it's that Trove should support alternate compute back ends and potentially interface directly with Magnum The other point of view was that Trove should continue to focus on a single compute back end Nova and not really deal with Magnum and We should use Nova drivers as a way to orchestrate bare metal virtual machines or containers And we had we had lots of discussion about this. We also Had some input from experts on on Magnum and containers and we decided to go with the Nova based approach We we also discussed a couple of projects that would significantly extend Trove's capabilities in the area of storage currently trove relies on cinder as the only storage back end and There's a project underway to extend this potentially to have Manila as another back end There's a project which will try and better use capabilities of storage like cinder to do things like Snapshots for backups and so on these projects are going on going ahead right now, but these were some of the Really interesting discussions, which we had at some In these discussions, you know, did you talk about specific user needs was a lot of feedback from the from the broader user community Yeah, we were we did talk about Specific user needs. We had we had some members of the user community who participated at the discussions And some of these are our specs which have been up for review So we've had users provide comments and reviews on the specs as well. So one of the things which And I have to say I'm Lucky to be able to build on some of the work done by the previous BPLs We have a fair amount of feedback from end users of trove and I'm lucky to have that feedback continue So if there are other people who watch these videos and are interested in databases Yeah, please do feel free to come and comment on our specs. We value your input great, that's That's exactly what we need to do across all the projects Yeah, so what what do you see as the Sort of the the top three priorities or news user features that you're going to roll out as part of the new release Sure, so There are so let me let me maybe answer that question by talking about A very significant problem which users brought up which we we need to address we talked about upgrades as one of the projects Which we want to work on architecturally trove has a A guest instance typically a noble instance. It may be bare metal. It may be a vm What have you which runs the database which the user wanted, but it also runs guest agent and running the guest agent in tenant context is always been a A challenge. So there's a very interesting project called superconductor Which proposes to address this by having a new controller side components, which will manage the guest instance remotely so To answer your question with the two priorities, which we have we want to as a project Continue to rapidly improve the capabilities that trove provides but at the same time While we add these new capabilities to control clusters and things like that We're looking to make sure that we're doing also maintain interoperability with old clouds with existing users for trove And to make sure that data is not lost while we go through this. So key projects. We're looking at additional capabilities for clustering additional capability for high availability through anti-affinity and affinity for placement of instances clustering for couch base back up and restore for postgres Improving our ci testing with scenario tests. All of these are projects, you know specific projects, which we're working on Which we'll try and get to that point where we improve the capabilities of trove make it more enterprise ready and better able to serve The the needs of people who are using trove Okay Would you say there any particular seams that you were trying to to to focus the development activities on for for newton? um, we've been trying to do this now for A couple of cycles the mitaka release. We consciously started to do it in mitaka Trove as a project grew very quickly. We accumulated a fair amount of technical debt But while we're doing that we've now come up to a point where we support almost a dozen databases and you know Clustering and replication and things like that. So in the mitaka release and now again in the newton release We're going to focus on things like interoperability stability ease of use And those three things are you know, if you will put if you put each of the projects which we're Looking to do it'll fall into one of those releases interoperability stability or ease of use Okay, and when you say ease of use is that End user focused or also operator focused in terms of ease of manageability So trove has two users two typical users Trove's one of the users of trove is the operator The operator may be a cloud service provider. It may be an it person The other user is the person who is provisioning the database. So we're talking about where the ease of use we're talking about both of them We're talking about ease of use from the operator perspective for example An operator may want the ability to say if you wish to provision a mysql database and you're in a development team You can only do that on instances of this size and you can only use traditional storage But if you're provisioning a database for production use you can use larger instances and you can use solid state traps so that's a That's a level of control which an operator needs and we're trying to make that possible for them And it's actually one of the projects which we are in fact working on The flavor support is already in the volume type support is something which We're hoping to get in for newton Okay, great And and if if I can I know you're probably hedged down just getting getting newton underway But what are you if you look out towards okada? What kinds of things do you think? You'll be looking at in that timeframe So several of the projects which I described to you are not going to be Completely delivered in newton. There's going to be superconductor as a project may take multiple releases to Get out to to users the project to deal with upgrades image upgrades or image-based upgrades is going to be a multi-release process We have people already using trove in production And therefore in order to get them to the point where they're able to fully use this capability will probably take a couple of releases So some of these projects aren't going to be Multi-release projects. I think the container effort is certainly going to be a multi-release effort Some of these are fairly large Bites which we've already taken When we started working on replication in juno and clustering in In kilo we realized that it was going to take multiple releases before we could support all databases with replication and clustering So those are some of the major projects which are already things which are ongoing which are going to go into akata as well Okay I think that was all the prepared questions I had is there anything else that you think it's important that users understand about The trove project and and where you're trying to drive it and You know any other feedback for for users at this point Sure, I think one of the things which I've I've been really heartened to see with with trove is From the ice house days to now You know the lower two years trove has evolved very rapidly and and Been able to deliver a significant basket of capabilities for users So if if somebody watching this video is a user of open stack and they have some database in their infrastructure They should definitely check trove out. It is relatively easy to get it up and running And you can very easily find that trove is a very very powerful addition to your existing open stack cloud Another thing which I will add is that trove is a rapidly evolving project and we're always looking for new contributors to the project So no matter what area you want to assist with we can absolutely use your help So please stop by either the irc channel open stack dash trove Check out a copy of the source tree Look at launch pad. There's a lot of bugs which are marked as low hanging fruit If you wish to contribute with documentation we could use that as well. So Literally anybody who wants to contribute trove There's something you can contribute you can either use a product or you can contribute to it for other people. So Come on over and join the party Great um, um, thanks for your time and and your insight