 all right okay so we can't talk about gossip anymore all right good afternoon everyone so we're just saying it's so glad to see so many of you showed up because we're at the the end of the day the end of the summit and you know I'm sure you know all of you must be pretty exhausted I am I've actually I've lost my voice and so glad that you are here and you know I guarantee this panel is worth your time we got this esteemed panelists here and you'll at least pick up a couple of new things yeah okay my name is Annie Lai let me introduce myself and I work for Huawei and one fun part of my job is I do travel a lot I travel around the world of you know ever since I joined Huawei I've been to like over 30 plus countries and all like five continents now talking to open-stack you know users and customers and one thing I learned is you know one of the reasons that you know customers are choosing open-set is interoperability because they think with interoperability there's no vendor lock-in because you know especially a lot of CTOs they have experienced the pain of being you know controlled under by one vendor or two vendors and now her average like CTOs job like life span for their position is like three years so they got smart they don't they want their architecture to be fully open and they want their solution to be you know fully interoperable so that's why we're here today to explore further you know the definition the meaning of interoperability and you know the promise that you will deliver so all right that's let me introduce you the panels and I think they are pretty famous you've probably know them already you know they are frequent speakers at the open-stack summits and first I have Brad who is from IBM is also open-side board director and later I'm gonna let them talk about their involvement in interoperability the subject and Agli also she's open-stack you know everybody knows her she's been here for a long time and she's also said she's been sitting on open-stack board for quite a few years now third year wow wow so that's I mean Agli has been doing tremendous work at the board level as well as at various in various work groups and we have Kurt who is from do a T systems and Kurt before their deployment of open-stack and you know you got involved even before the deployment of Deutsche Telecom OTC open telecom cloud so he definitely has experienced the process of this whole interoperability so later I'm gonna talk ask him some tough questions about you know his experience about interoperability so okay so I have said enough and first my first question for the panelists is what's your role in interoperability at work in the community what's your role okay so I'm Brad I've for sort of two releases now but me the overall lead of the interoperability challenge you know it focuses what we see up on the keynotes and it's a very holistic approach to interoperability you know there's something special about getting everybody up there and having to run the same stuff with based on the same scripts really helps to make interoperability a pragmatic discussion as opposed to a philosophical or theoretical one you got to get everybody up there and they're gonna have to do their things so I've been involved from that from that portion also the the Ptl for ref stack which is the interoperability project we'll talk more about that later Catherine Diaz who's in the audience there works for me as well okay so my name is Agnes Ziegler and I work for Act Space I have been an open stack board for now it's my third year and I am a co-chair of interoperability working group if you have heard of deaf core committee before it's the same thing we just renamed it to be maybe easier to understand what interoperability committee for working group focuses on is defining standards or guidelines for what it means to run open stack so if you are working on a product that calls itself open stack private cloud or open stack distro or open stack something else you have to go to the foundation and the acquired license to use the logo so one thing that foundation will ask you is have you passed the interoperability guidelines and we're the ones that actually govern the guidelines so the working group is governed from the bi open stack foundation board but we also need community input so we also have a elected co-chair from community and anyone is welcome to get involved in our working group and help us with with those standards or guidelines right now the guidelines are more like of a spec just because you pass the guideline doesn't mean that you have fully functional open stacks a lot more features that we're not including in the spec but that's a different discussion okay let's talk about those later and you Kurt yeah well first let me confirm that's the interoperability group is very open welcoming to new people so I've experienced this myself oh you sign up good we need help we need lots of help so no but I'm yeah I'm attending and joining already since it's a bit of time and I really felt welcome from the first day so that was good and interoperability actually is an important topic for us because a lot of the discussions we have with our customers they tell us well we have chosen your platform open telecom cloud because we did not want to have this lock-in effect and they asked us well how can you show us improve us that actually that is the case that works and then this open stack powered and into our program is really an important piece of the answer to that question and that's why it's important to our customers that's why important to us that's why we participate in that community can you share with the audience what's DTOTC do it open telecom cloud what exactly is that because some people here might not know what even though they should no thanks for asking I mean we launched in public cloud based on open stack more than a year ago it's a global cloud but we started in Germany which is where the largest part of our customer base is and one of the USP's that we do have is to be able to fulfill the European data protection and data privacy and security concerns and laws and that is one of the things that has prevented some of our customers to use the cloud so far and now they have that possibility so it's one of the things we do there perfect great and I want to go back to Brett what Brett said interoperability is a big word overloaded word and everybody has some idea of what interoperability is and the discussion sometimes can get to a pretty philosophical so let's just talk about that you know you know more idealistic way what's your vision of what interoperability should be you know in the end that customers can truly benefit the promise of interoperability right so you need to be a little bit careful you know typically as you go down a process of you know in the open source community to drive interoperability you're gonna start with things like ref stack where you're defining you know what what needs to be done and what everybody needs to implement and there'll be certification tests and people will do that but we have to remember the true test for interoperability is what the user experience is so if the things that cause interoperability problems are the choice of automation for how they deploy and provision for example not all the cloud support open stack heat not all the clouds support terraform that customer or user or operator experience is gonna be hey this stuff is interoperable I went and I was doing my stuff with heat over here and then I went to this person's cloud and it didn't work they don't want to hear that you're gonna say yeah but both places passed all the tests right so we have to remember it's an end-to-end experience of can we make it incredibly simple for an operator or a user to say you know and this is how we like to say it from a marketing perspective you know you know Brad we're using your cloud and God forbid you decide that you want to stop using my IBM cloud at least you can easily move to another vendor and you can go to Annie's cloud or Eglise cloud and yeah the same and since that's okay I can maybe then go and you know sell you services anyway because we're all on the same platform but the nice thing to you as a customer operator is that freedom from vendor lock-in you know we used to have these stories that you know if you go to a proprietary cloud or what have you in the 1970s we had a little commercial called the the Roche motel where their roaches would check in but never check out you like you to be able to you know God forbid I don't make you happy you should go check out you should go to Egley you go to you know Annie whoever so that holistic approach that's the proof is in the pudding and the me showing up with these little spreadsheets and check with saying but we both pass that just doesn't count from from a user what about Kurt you're a large user right and the interoperable you say it's very important to you so for DTOTC what would be the true vision of interoperability that's going to help you differentiate DTOTC from like Amazon and these guys well you bring up Amazon so I mean Amazon has this world domination piece I mean they are building this huge cloud usually successful and I mean a lot of the things they do to do well there's no point not acknowledging that but I think the open-stack community has a different vision the vision having many clouds smaller clouds that somehow can connect together and together be stronger private clouds public clouds really allowing customers to choose what best fits their needs and then to relatively easily move from one to the other and use some combination of these and expect that to be a seamless experience and that's kind of the division we we wanted to build as an open-stack community and which we still have to work on a bit more to get there right you know I also have a perspective because you know Huawei we also enable public cloud private clouds and I think you know one of the thing that's going to set us apart from all those proprietary cloud platform is our interoperability value proposition so you know we have you know right head Ubuntu you know suzai and then all these you know private clouds and then we have DTOTC customers environment especially enterprise customer environments are going to be hybrid so they're gonna have you know specific usage for private clouds specific usage for public cloud if all these clouds are open-stack powered and later actually was going to describe what that means and then you know they can be interoperable then that means workloads can be you know migrated you know back and forth can run back and forth you know to me various requirements and we can do that and you know and I think that makes open-stack's value proposition a lot stronger that's my perspective so so going back to Ackley and so so from what I just said you know with the current open-stack branding like open-stack powered open-stack compatible you know we'll have all these trademarks and can you give us some background about why we need those trademarks and what they mean and how they can help us achieve the ultimate vision of interoperability so first of all Brad is very right that yes there is a checklist but just because you passed the checklist doesn't mean that it's going to be perfect so what the foundation has done are basically we help them create a bare minimum to achieve the certification to so right now if you go to the foundation and you say I want to call my product open-stack cloud you have three options you can either get open-stack powered platform open-stack powered compute or open-stack powered storage platform covers both storage and compute compute is Nova Keystone Cinder Neutron it does not include Swift storage component is Keystone plus Swift so what that means if is if you're running all these main projects you can get the open-stack powered platform logo if you're running Nova Keystone Neutron and Cinder and Ceph you can get only compute because Ceph is not an open-stack project and it does not cover the open-stack storage component even though you may be able to pass the APIs for the storage component because you're not running open-stack code you're not you're not going to get the open-stack powered platform logo just because you're not running all of the code and that's the very important distinction API passing APIs is not enough we do we do expect that you are running open-stack code not just another re-implementation of the APIs I think I covered all the main aspects right now we don't have a compatible program and you just get the three different you could get three different logos oh I think there is a training program but we're we're not concerned with that in our interoperability working group we are talking about adding different programs one of them is add-on and another one is vertical add-on would be something like if you want if you are running additional components like designate and you want to say differentiate your cloud and say hey I have open-stack with DNS so right now we're working on creating this the vertical program would be something like NFV ready open-stack we're working with open NFV community to figure out what that would look like and what that would mean so but that that's that's in the future well that's very much in you know alignment with what Mark was saying you know and doing his keynote is composable architecture open-stack is a composable architecture so the users can you know pick and choose whatever building blocks they need to help them you know meet the requirements of various use cases so that's awesome so open-stack so what why do you think it's important to get that open-stack power and open-stack while compatible is not to know let's just focus on power trademark what why do you know I know the community manager who is responsible for all these you know tests are this Chris Hodge right he's constantly pushing people hey you're gonna need to get certified you know so you can earn the right to have the to use the open-stack power trademark and why is that so important for vendors and service providers and users I think it's probably most important for users because as a user you may assume that anybody that calls themselves open-stack are actually running the same thing when in fact those of you that have ever install open-stack you know that you can fine-tune every aspect of each project I think just a policy file for Nova alone you have over 400 different options so you can turn everything off or turn everything on for everyone and everything in between so as a user you don't know that or you don't want to know that and you don't want to care about it and you're like I just give me something that creates VMs gets IPs have some networking maybe some storage and so just some basic cloud things so that's what the logo program is meant to help with so if someone calls themselves open-stack that they at least have the bare bare minimum where they can go and create VMs and get some networking and some storage for from operators perspective I think it encourages them not to go and create too too many customizations if you create something custom then you also have to maintain it and if you get too far away from trunk or mainstream open stacked and you're not going to have an easy time in six months or a year when you need to upgrade yeah I just want to say you know actually has done and interrupt world group they have done tremendous work it's a lot of work to you know push all these programs and so I just want to give a shout out they are doing amazing job and just to help you make your life easier can you share with the audience that what kind of help you need because we don't want you to be losing your sleep and working on this all the time so get help yes if you are passionate about a particular project and you think we should be an add-on for example DNS database heat orchestration and you really want to see that come to life sooner rather than layer what come and tell us about it and we'll say we'll we'll let you know how you can help us if you are passionate about any other aspect about open stack same thing we can probably use her help there interoperability working group is really a lot about documentation and procedure and process we do need help writing additional tempest tests if that's something that you are interested in will you tell us we'll let you know what kind of test we need help with but a lot of it a lot of the work is not very glorious it's documentation and process and making sure that we're testing the right things and that the program is doing what it's supposed to do that we are protecting the open stack yeah trademark and that people that have that trademark actually have open stack and sometimes the balancing act because if you if you're running a private cloud and any is running public cloud well the requirements will be different so that's why right now we're just bare minimum that everybody agrees on right so you still so there's still a way to make this whole program you know to me the requirements of interoperability there's still a way there so please you know help volunteer and participate so from the other side of coin from a user or big operator standpoint or you went through the pain of getting your DTOTC certified or you know now your DTOTC is open-stack powered you know platform and can you share with the audience the experience of you know going through that whole process and can you can you tell us give some as to give us some tips on you know for any kind in any company or any user wants to go through that certification process what kind of things they you know they need to be aware of and you know so they don't have to go through the pain that you went through yeah after all I don't think it was that painful but okay well fortunately but I mean indeed we started before we launched because I wanted to have kind of make sure we're doing the right thing and I wanted to make sure we have kind of a baseline test that tells us about the quality of our platform and I wanted to make sure we push the right people the right way or when they're ourselves and then as soon as we had launched actually had a lot of customer discussions and then they were the ones pushing us to make sure we have a good answer that question to the question but yeah I mean what we did is actually we we put two people on the job for time and they set up a bit of infrastructure we actually run our own for for some reason that we initially had to find a way to kind of limit down the list of tests we perform and that always have to hold because we knew some of them would fail and then screw up the environment that can be some some complications we did all that and then I think that was like kind of the the initial investment that we had to do to learn how it works and to interpret and to learn to understand okay there are some duplicates because there's some name aliasing and stuff make sure we understand accounting to measure how far we away from the final results and then I think we pushed a bit of work on our part indeed we had to fix a few bugs we endure the pain for you I'm glad you feel painless that's our job okay so so that's good so now I'm going to back to Brad so you were talking about you know in the interoperability challenge and you know I'm not many people not everybody knows I'm sure you have seen the demo and stuff we'd like to have an understanding of what happened in the background how you made you know how you pulled all these companies together and you know do the test and run the test and do the life demo and you know what purpose does it serve right so just going back to the beginning because it's fun history this was open-stack summit Austin my IBM general manager Don Rippert had a presentation and he made it clear as part of the presentation that the issues the open-stack community were facing that he was seeing when we talked to customers and analysts and and that the two big things originally that that that that we were getting dinged on was show us you can really run enterprise applications and show us that that inner open-stack is truly interoperable so as part of his presentation he threw down the gauntlet in Austin and said I think we should have an interoperability challenge who's with me right and then he collected all the cards and so the funny thing was you know so you know you have your general manager gets up there and collects all the cards and you're sitting there as the distinguished engineer going wow I bet this is gonna roll down this work for me and so what we did and how we did this myself and my boss Todd Moore and way back when is we went into all the open-stack Foundation board members and we said okay well so-and-so from your company gave us our card and and Don collected all the cards and we want to have this interoperability challenge and show that things are interoperable and so we wouldn't want all the board members and everybody got buy-in and what you do is you say okay when only buy-in I need a technical lead so tell me the person who's your technical lead so you have to do this this round of management to go around and say okay we're getting all these technical leads from all these folks we're gonna participate and what we started doing is okay well we need a really nice enterprise workload this is way back when right we're getting ready for Barcelona we need a nice enterprise workload well I don't really see one and then you look for somebody like well who do we think can build a really nice enterprise workload and the name that popped into my head was a guy named Tom Lee he's sitting right there and Tom is a hero he built us a lamp stack application he also did some work we had some folks in IBM who built a Docker swarm app and whenever you're trying to do something across a community in a perfect world everybody contributes evenly on the hard part kind of things but the reality is these are all very busy people doing part-time things and you need those folks that are gonna really lose sleep over this all coming together and can serve as rovers and help and you know that's a lot of things that Tong did helping all the different folks trying different things out and we learned as we went everybody a lot of great ideas well we think we ought to use heat for for automated deployment well we think we ought to use juju charms well we think we ought to use terraform well we think we ought to use Ansible and the beauty of that environment was you know the guns to your head what's gonna work Ansible work just just let you know you know let me ruin the surprise Ansible is the thing that worked and you know Tom will look at me go Ansible keeps being the thing that works Brad we got to keep using that and so you know getting ready for that first one in Barcelona we learned a lot of lessons we got everything coming out and we got everybody up on stage in Barcelona for the first time that was a lot of pressure my general manager was on stage six feet behind me and we've never done this before and when I think about it who what kind of crazy person puts 16 different clouds with 16 different networks and says let's put them up on the big screen and see who's running so we did all that that went great and then we moved into this phase two of that I talked about at the thing I don't want to get too much time but then the phase two of moving into well let's look at Kubernetes multi-cloud cockroach DB and and all the fun in between with all that that's awesome so we want to thank Brad and your team for such a tremendous effort because you know this is definitely it's an important milestone of our interoperability and we got a lot of press out of the two demos that we had that was impressive so thank you and also thank I want to thank Agli and them and then Kurt for participating absolutely they both participated both times right yeah so that's awesome and then Tony too I mean he's the hero who worked behind the scenes so yeah so anyway so that's what communities about right we'll have to chip in and that's good and so I have a few more questions I'd like to open up for the audience to ask questions do you have any specific question about interoperability and these guys are all interoperability experts so this is an awesome opportunity to ask them questions in this subject and please go to the mic if you have any questions so because we're being recorded no question are you interoperable yeah all this one okay I have a question Brad is going to appreciate it so now people doing putting arm processors and other type of processors in clouds so you're talking about interoperability and for me you you're just like assuming everything runs Intel what about interoperability across different types of CPUs so actually this this time during the challenge I wouldn't have anyone running on a different architecture but last time in Barcelona we did it was someone from Lenaro running on an arm processor so it worked just fine and if you watch the you know you'll hear Gemma talk about it about the processor but right and so arm did work but some processors have issues and you know that's the beauty of you want to be up on the stage you're gonna have to find a way to make things work and so those issues can pop up go use a mic actually I just want to add a little bit to that question at the launch time actually I was just having a sitting with a guy from Cray and he told me that they run OpenStack on their system so I encourage him to participate next time the intro because Cray is a little bit different architect but I mean from what he told me already that they're running OpenStack on Cray that's awesome great and the IBM has open sec running on open power correct yeah I mean potential for next time yeah yeah that's awesome any more question if not I have one question my last question for the panel so you know so we talked a lot about you know the vision of interoperability in the beginning and and then we're definitely putting a lot of efforts to kind of push our community to work towards that yeah but moving forward as a community whether you know you're a developer or user what should we do and or what should we be thinking about in order to reach the you know the owner of interoperability for you know our users they can truly we can truly say interoperability is the biggest differentiator well now because it is a huge differentiator of OpenStack you know while it's being compared against all those proprietary platforms yeah so I mean I'd like to see you know a catalog of just a variety of workloads that we can just feel comfortable that run and you know maybe there'll be slightly different profiles you know we also did a lot of work on the NFV side and we didn't talk a lot about it but there was a different session on it but it would be great you know Nirvana is you can go and see you know the proof is in the pudding you know here are the 10 to 15 workloads that that we're confident are going to run around all these different platforms they're in a variety of different domains here are the ones that are for NFV here's the ones that are just general you know OpenStack and having that confidence and people being able to look at the best practices everything's done in the open and these and these challenges you know you watch our progress in our IRC meetings you want to watch our progress and the repositories and you know what's been submitted to the Ansible Scripts and what's been working and ideally you know everybody could really gain and benefit from those best practices from from all those different workloads yeah and I'd like to add to that if you are operator do what Kirk did first look what is required to pass interoperability guidelines it's not a whole lot I think total it's that you have to pass about 200 tests API tests in total so that's not a whole bunch so please be sure that your product can pass at least that if you are a developer also please take a look at what's there and don't try to create something completely different I think right now we're facing challenges with Octavia and Elbaz having slightly different APIs or slightly different endpoints so don't don't try to be too creative you know whether you're operator developer follow standards follow guidelines and you know meet those and then build on top of that don't try to deviate from that that's not going to make your product special it's going to give you a headache they run when you're trying to pass the guidelines and if you are having issues with the guidelines come talk to us we do make mistakes sometimes we need to flag or remove a particular guideline and that's fine you know don't assume that we're unapproachable we definitely will work with you try to figure out what's going on whether you have a good reason to implement something the way you wanted to for a good reason or if it's just something that well it's probably not a good practice I can also confirm that thank you thank you do you have anything to add yeah I guess resisting the temptation to differentiate in the wrong areas is something that is very important that each of us needs to remember some when those are have learned over the time to to to understand this better than others and I think we all as a community grow taking it maybe a step further I mean we are covering a baseline with the interop working group entity the power program and of course the next steps you described they will help us to actually make this make life quite a bit better with the vertical programs and the add-ons we need to we need to make those successful and put them interaction but I also think we will we will need to go even further with that to to be able to actually live up to the vision of having like federated clouds that can work together and look seamless to the customers and I mean there's things like discoverability which we have not really covered in the interop program other things we need to we need to tackle to to to be successful there I agree right now the current programs they they're not meant to be they're just meant to be the very minimum you can add anything you want on top of that first as long as the minimum as long as you meet the minimum you can implement all of the projects expose everything to your customers don't let that stop you perfect thank you so much I think I hope that was informative and I'd like to you know thank our esteemed panelists and thank you so much for sharing so much valuable knowledge with us and I'd like to thank the you know audience for your participation thank you