 Hello everybody. Thank you for coming to the session Which will be a round table. We have different panelists that will introduce themselves. So the question we want to ask to the panelists and We hope that you will love their answers Is around containers and all their representing future of infrastructure as a service And how does it position themselves with regard to other technologies? And so for this round table we have Maybe maybe maybe it will come we have four three panelists Simona on the left hand side Jerome in the middle and David on the right hand side For me. So it's a reverse for you, of course And they will introduce themselves coming from different angles working on different technologies So they will introduce themselves and then we will ask questions and if you want to make the Discussion interactive feel free also to raise your hand and ask questions And I will repeat them so that you can be recorded and the panelists will will give you answers to those questions Simona, can you stop? Yeah, sure. It's working. Okay So hi, everyone. I'm actually really excited to be here and I want to take Bruno as well for the invitation I work as a product manager for Suze and I will just take one minute to introduce Suze I would love that everyone knows or let me first ask. Do you know what Suze is? Okay So I work at Suze as a product manager I have been doing product management in different areas from the maintenance side of the classical operating system moving into Developing a service pack so a version of the operating system and now building a platform for containers And that's the primary reason why why I'm here to talk about containers in our vision for containers Hi, so I'm Jerome Pittazzoni and I work for Docker I do mostly evangelism these days So you might I've seen me in how to get your microservices working on containers without blowing up everything Things that I'm not doing anymore But Brett and Laura that were here in the room are doing a workshop about that So I recommend you to check it out if that's the kind of thing you want to do But before doing that I was doing Ops at dot cloud the company that became darker and before that I was doing mostly development And I feel like this kind of Discussion for me is a kind of back to the future because almost 12 years ago I was trying to build infrastructure as a service back in France in the small companies They had and we are also looking at that thing Everything like UML and not the the thing to build your diagram But user mode Linux, which was a kind of ancestor of containers So I feel like we might have a shift for circular here David hello, I am David Flanders of the OpenStack foundation So I primarily work on the container and platform as a services that get connected to the OpenStack API So I work a lot with SDKs and different developer communities from lots of different programming languages trying to understand how they want to connect this plethora of Container choice that we have with all the different Stateless features and tools out there last I saw there's about 70 Gartner's tracking about 70 different Container and platform different services and modules and things like that. So it's a very confusing space And yet I think that we're obviously going to see a lot of Filters going on where we're going to need to make some choices I'm a big believer in the fact that I think we need to move towards a kind of lamp stack for the cloud And so I'm I'm genuinely excited to be chatting with Simone and Jerome on how we do integrate all of these Technologies and how we start to build this new lamp stack for the cloud So we have the true open source that we want to be able to build the things that that we love building with the computer in the web Okay, so my first first question for for you would be around the Solvable tools that you need to build that stack. So We have bare metal systems. We have virtual machines and we have containers now How do you position them? How do you see the useful mess with regards to each other? Where are the key points for you around those technologies and maybe David you can Sure, so I think the most important thing to remember here is that we do work in a space that does change a lot So compute modalities We'll we have very old compute modellies. You heard Linus talk about and and Alan head of Seuss talk about mainframes, you know So I worked in universities for a long time and you know, there are mainframes still kicking around Likewise, we still have virtual machines and a lot of you know enterprise Legacy applications running on those Bare metal especially as it relates to things like GPUs and being able to use new compute modalities are coming into play more and more to Experiment with and of course containers And a lot of people, you know, think open stack doesn't love containers We love containers and containers live open stack. We we are building open stack better with containers from just a point of view of being able to update it to keep it Modular and being able to allow more people to use it and likewise. We're putting caner containers on top And I'll talk a little bit more about that later But I also and I just came from a conference on edge computing where we are starting to talk about all the new compute modalities coming down the line, you know things like Why not array computing and being able to do that smaller compute modalities things like serverless and functional computing and then also that looking kind of in the far future I think we we've got quantum computing IBM's just launched, you know a cloud that you can actually go and start using Cubits right now and start playing with quantum. So quantum's here So I think we're going to continue to see these new compute modalities for all the use cases that we see and we're just getting More and more use cases, you know, it is getting bigger and bigger So for me, I'm I'm actually genuinely excited these past five years have been a lot of just making virtualization work and making containers work and making bare metals work And we're starting to see a whole new set of compute modalities open up. So I'm I'm genuinely excited about that I think that generally everyone is really excited about this now What we are seeing with now on our customer base is that everything is going to be a journey So we should not expect that even if the technology is extremely cool And it's proven to be working and bringing all the benefits like improving Productivity bringing the innovation back to the organization and so on for some organization It will take a while to get and build their container ecosystem and container architecture and You know, we started we are talking about infrastructure as a service And I would like to maybe do a little bit of detour to talk about how we ended up with containers and everything So I think that all of you in the room already know that You know managing an infrastructure It's usually a very risky task and it brings a lot of responsibility to assist admin New concept like continuous integration. They sounded really cool But those can be really dangerous when you use them on mission critical infrastructure And the reason for that is because your infrastructure, which is critical for your business might not be available Once you do a patch and you cannot have the mechanism ready to roll to roll back easier That's why and we are seeing that on the operating system We know we are always exciting about new versions of the operating system But large organization and enter pricing running Thousands of servers are not going to roll out our New York's version that in the immediate next day after a release They will usually do it gradually with a lot of testing making sure that the critical infrastructure It's still up and running now historically application have been linked with the infrastructure and historically at the application level we have Inherit all those rules on the infrastructure And I I bet to say that in the last 20 years We tried to decouple the application and they allow the application to be more rapid and develop itself Embrace the agility the continuous integration and everything while the infrastructure on the infrastructure level We continue to keep it stable and reliable. That's why we still have this usually at Susan when I talk about DevOps We try to explain to everyone why this is usually a fight between Operation and developers and why we want to embrace them as a as a core concept Now when we look at what's happening now in the market and this is what do it open stack and so on what happened Is the innovation needed to hit the infrastructure and we needed to find a way to automate and to bring all this Application innovation that we had build around with agility and so on on the infrastructure level Allow the CIS admin to have tools and automation and stable solution They can rely on so that they can get the same agility as the developers have on the other hand and You know when we talk about cloud and application in the cloud everything that could be an easy fruit Everything that could be easily put into the cloud as an either as a container or the virtual machine I bet to say it's already there if you have a small website your marketing campaign stuff like this are already in the cloud But what we are seeing now is that organizations? They they want to continue this this journey and they are looking at the what we call more enterprises or their mission critical Application and today is not easy to put legacy building system in a in a cloud and not even talk about containers Some find the easy solution of just checking a box saying here is my container strategy I take my legacy Oracle system. I put it into a container I put this to the container in a VMware. I running on top of open stack and I have the task done Yes, it's true. It is a way of doing it But you don't yet gain the full agility that micro services Approach or a modular approach will bring you so that's what we are seeing today that customers are struggling on defining their architecture of how to embrace containers on top of infrastructure as a service So for us at Susie infrastructure as a services core to the future of the software defined data center Allen has shown the the slide. We usually have it on all our presentation We have at the as the first pillow the operating system everything We still need an operating system to run everything but the second pillow It's an infrastructure as a service and then we go up the stack with containers with Kubernetes with Cloud Foundry And then you get a solution or you get different solutions depending on your needs So that will be in a couple of words our vision So as you can see we kind of all agree that the future of containers and IAS are deeply intertwined So the I feel like the the good question would be more about how like how how do we get there? And what are the things we can do? to to to get some interesting results and I think one one observe interesting observation is to think about the fact that the biggest and most successful Public IAS infrastructures are built on top of containers if you take Google Cloud It's when you when you start a VM on Google Cloud It's starting on on top of something that looks like a container that is using C Groups and other mechanisms that are the same mechanisms used by Containers to make sure that this container only has access to a specific amount of resources there is an orchestrator which is Borg which is the kind of Ancestor of Kubernetes So Google Cloud is built on top is running on top of containers, which is kind of The reverse of what we could think at first is like oh, I'm going to run my containers in VMs Obviously, but these people are running VMs in containers. That's interesting And if we look at another huge successful public cloud Amazon It's it's not running on top of containers per se, but when you start a VM You indicate a machine image which is really close to a container image And whenever you want to start a VM you always have to specify that image and those images can be made from Running VMs so you you create them the same way that you create container images kind of so the the lifecycle of those VMs Is really close to the lifecycle of containers, especially if you want to do it right and by do it right I mean in the like DevOps ways immutable infrastructure, etc. etc. So Ias and containers that I think at some point will become almost impossible to distinguish in some way Because that the perfect Scenario if I if I wanted to deploy something tomorrow I wouldn't I don't want to choose between containers and Ias I want both because I have some legacy work workloads that run in VM and Sure, I could put them in containers But that would just be lipstick on the pig that thing is still a VM And I also have new workflows that are containers and I would like to address both workflows from the single control plane and and see them in a kind of homogeneous way And the last thing that will Probably at least some reaction from my neighbor will be about as an open stack For instance as this reputation of being complicated to operate Kubernetes is getting there as well And so it would be interesting to see if we can take some really simple Container primitives for instance Swarm as the reputation of being really simple to to deploy and now see okay. Can we layer? Something like open stack or some kind of Ias on top of that. Can we get either a simple way to? Deploy and operate open stack thanks to those low-level container orchestration mechanisms Or can we get an even simpler Ias thing? That will let me start VMs as easily as I start container by doing like talker run blah blah and then 10 seconds later. I have a VM up and running on my infrastructure So it I actually really agree with this We're playing with all of this right now at open stack and what's really getting excited What's happened with kubernetes and with docker and with those 70 other companies out there who are now doing container solutions Is that we're getting a lot of experimentation? It's kind of like the the Cretaceous period of speciation inside of the way you build your applications So things I've just seen already and drums actually had a great example Borg Still using VMs with containers Amazon still using VM containers We're mixing and matching and there is that kind of blending going where you almost can't tell the difference right because there are things in VMs which are really brilliant some of the old security layer stuff that goes inside of there and all the rest of that And there are brilliant things for the statelessness and what you want to achieve with containers So I think actually Simone you hit you hit the nail on the head as well Is it's about walking people through that journey as we get there now? If you want to check out some cool stuff that is starting to happen with this blending I've seen for example open stack cinder, which is our project for being able to do a database back ends or sorry Not database, but storage back ends Cinder has been working for seven years now And they've been working on actually creating over 80 different driver profiles so that when vendors come up with new compute You want to sell it? You can just grab cinder put swarm on top off you go And that's incredible for me to be just be able to do that kind of lightweight experimentation All of a sudden you're seeing web companies again who are able to build their own little clouds Just by using a couple of modules and things like that They don't want to just use utility computing They actually want to still have their own things so that they can touch and work with it So we are getting to an arrow now and again I go back to this kind of lamp stack example where you know Linux doing all the work Linus. What is he doing day on and day out? He's making sure that drivers plug in so that we can have more computers utilizing this stuff broadening the scope Enabling more people that's exactly what open stack is doing you look at cinder Swift Nova glance all of those are trying to work with hardware vendors to make sure we can plug the hardware stuff in so that You can use these things and then as we get into you know the Apache You know back in the day It was the Apache web server Docker and Kubernetes and core OS and tectonic and all of these companies They're starting to provide that interoperable Apache web service thing. We still got a ways to go You know like what does scripting on the web look like you know I've been trying to keep my own serverless and functional Maybe that'll kind of fulfill the P in the lamp stack You know when will we get to a place where databases will be you know truly serverless you there's some brilliant Cockroach DB I think is doing some really cool stuff now around this. I'm sure Docker has got some things as well so My advice to anyone on on about this kind of thinking about I ask versus containers is it's not an or It's an and just like the lamp stack. You had a bunch of different foundations You had Apache in there you had Linux in there you had my SQL you had all the different You know programming languages all of those are separate groups separate foundations. They're not gonna be consumed by one company We don't want them to be consumed by one company or just one foundation We want them all working together in correspondence. So just remember that when you're kind of going out there and people saying are you using Kubernetes? Or are you using? Docker or are you using open stack? No, it's an and it's an and we're using bits and pieces from all of this And most importantly like good engineers. We're letting the best technologies actually prove themselves So so what will make the customer choose one technology versus another is it typically is a type of applications are developing So general mentions a lipstick effect on on on traditional application used in container mode So do you think that you really need to tie containers to cloud native type of applications and that? Legacification will never benefit from containers or I mean you have a Docker You have the MTA program that you introduced so so some people at least a doctor think to them seem to think that it's possible to Build legacy application with containers. So So the the MTA program is super interesting at first when I was speech to program I was like what what is this this that doesn't make sense to me and and then I remembered a very important moment when I was neutralizing stuff for customers maybe ten years ago and One day I was tasked with virtualizing some I think an anti-server or something like that The hardware was ten years old and it was about to explode and I'm like, okay, this machine dies We lose. I don't know if it was Accounting or time off or so something important for the company I'm like, okay we need to virtualize that just to save that machine because it's in it's impossible to to get that running on a new hardware and I'm not really a Windows kind of person Okay, I mean for the whole weekend and then one of my friends told me oh, you just should try this software I can't remember the name. It was a p2v thing. And oh, what what is that? Oh, that's a physical to virtual you just run that on your server And it's going to work for a few hours Maybe days if it's really slow and then it will just like speed out a nice VMDK or whatever The VM image that you can run on your thing. I'm like that seems like science fiction to me And I don't think that will work But for now, this is my best bet to get that anti stuff running on Xen I had I think and It worked and I I was like well I I'm glad I didn't say anything stupid like I will eat my own shoes if that thing works And and so I think The what we're trying to do with the MTA program is something similar So at first it's like wait a minute when I'm counter-rising an application. I should carefully write a docker file and Like optimize the way I lay out the instructions, etc. Right, etc. Yes That's when you do it the cloud native way But you can also take an existing VM and you can like get that VM in a container image And of course you won't get like this super nice docker file that rebuilds everything in an optimized way, etc But it's getting there. The first step is just get that container and put it in the VM And oh, sorry do it the way around take the VM and cram it in the container and then little by little We improve the tooling so that in in the process we can get either a docker file of the equivalent recipe to be closer and closer to What a good professional artisan would do if they were docker rising in the first place so I I'm glad that I had this experience almost educated go about virtualizing physical machines because it gives me trust in the fact that Trust and faith in the fact that yes, we are going to take those VMs And we are going to put them in containers and some of them We will just like it's a kind of fire and forget thing now The VM is in a container and we will never touch it again because it's been running for years and years and We have no reason to rebuild that thing ever and sometimes we're like, okay, we need to like make this application like to Maintain and improve it over time and so we will progressively transform that VM into a proper Container, but so yeah, I think that's the one one possible path in this journey And I think what's important is to define what is a legacy application and this is what you mentioned So sometimes for me for some organization It is much cheaper to just continue to support legacy pay a vendor Than to look at this legacy and touch it touching it means I need to rebuild the knowledge I need to check the documentation. I need to realize ops It wasn't documented for the last 14 years So it's much easier to say, okay This is the amount I'm paying for this application to this vendor and it's going to be supported that day I can decommissioned and in my experience that's also this is the happy case If you are in the setup where you can say I will support this application for the next seven years And then it's naturally decommissioned because we only have a couple of customers or something like this What do you see more and more and more often is you do have a legacy application that it's actually one of your mission Critical application like a building system like something that is core to the business and there even if we call it legacy It still requires enhancements It still requires work on and one of the things that you are looking at at the moment is how do I Balance those two has or how do I maintain this application up and running because they cannot just you know shut it down The business is going to go down the next day And how do I move this into a more? You know cloud-native friendly setup and this is the I will say the new type of legacy Where containers can help and where I agree with Jeremy that it's a nice way to use containers as the first step So and look I think it I couldn't agree more with this, you know Developers we naturally get in this debate about what's better the VM or the container or you know functional or serverless and we get We get obsessed with these things because it does affect us and it emotionally hurts us when you know the thing breaks and all the rest of it, but really smart smart organizations have smart smart product managers like Samona Who actually understand that journey and the cost and that's that's one of the real things that I think we We can't forget about the most is that What is containers really pushing out there as a value proposition and for me it's portability of application and So having someone who understands okay We do you know this this is just an enterprise thing and it's used internally inside the organization and People access it via the one way and we're not too concerned about it It's when we actually start to think about containers from the point of view of in the future I want our organization to be cost ready to be able to jump around to the different clouds that we'll need and How we'll want them and that that's what we should all really care about if you're here at the open source conference And you kind of feel in the the economics of open source the economics says that we should actually be able to take our Applications and move them around and not just be stuck in one platform So more than anything else. I feel like we should be When you come to the decision points of should I be just throw it in the VM and use some use some magic to be Just get it in the VM and use it versus put in the additional cost and time and developer time to make sure that you're you're moving towards us a Stateless application which containers can provide if you get all the components in though. It's very difficult. It's time It takes a lot of time it takes very smart to developers to do it But if you can get it into that container portability Then all of a sudden you're starting to make your organization cost-proof you're able to say oh well today We're using Amazon and now we're gonna add Azure as another infrastructure provider and yeah We we actually do want to have open stack also as a private cloud there because it holds a bunch of sensitive secret financial data And we we really cannot allow that to be to be Put out in any other organization Maybe because of data sovereignty inside of a country or because of security reasons So let's not forget about the the actual economics of containers and the fact that it's about portability And enabling us all to have a better economic open-source future So I had one question back for you see my night So at Susie you had all the offering you had you have an open stack based solution You have your cast solution you have your past solution How do you advise customers with regards to all those different options? What do you where you tell them you should choose that type of product rather than that type of product based on the technology On the application on the life cycle of the application Well usually talk about their project and their life and their ultimate goals and it's all a matter of choice So they are customers who really want to start a journey and they are even satisfied today with only the container engine And we have that offering as well. So they only say I want to give it a try I'm just going to put a couple of containers here And that is the operating system plus the container engine then we have customers who are looking into the transformation That's why I was talking about this infrastructure transformation and then they are evaluating open stack And we are really proud to be one of the first open stack distribution and contributing to the open stack solution But nowadays we have customers who are coming back and saying, you know, everything is really complex And I used to get an operating a stable reliable enterprise operating system from you guys And now I actually have to deal with all okay What exactly do I want and so on those are usually the customers who are looking for a full platform as a service We're already the the log management. It's set up the monitoring tool It's embedded and you know the the task that it's on the sysadmin or that the task list for the sysadmin is really reduced Because this all comes as embedded full Solution provided by Susan that's the value of a platform as a service or how we call it nowadays It's a cloud native application platform. This will embrace this will Embrace the latest technology this will help customers run on Different solutions so they can run it on the public cloud or they can decide it to build it together with yes On their own private cloud and it all depends on the cut the ultimate customer needs and how they want to get there We are aware that more than 50% of containers today are running in public clouds So it's still a journey of you know transitioning from A standard for legacy usually application and set up to the new cloud native solution And one thing that we realized and that's why we have probably we have a unique set up Where we have a cost in the past and we don't want to play around in the terms and saying that the cost is the past and so on We have identified the need in the market to have Automation for containers and this is where we build it the container as a service platform and on top of this We are going to have the platform as a service the main difference is a container as a service is really focusing only on containers And it comes with very minimal that you need in order to run your containers where a platform as a service is focusing on your Developers and comes with all the tools that your developers need in order to run their application And you no longer have to worry about what is happening underneath. That's something that the vendor like Susie will do for you Okay Show me any feedback on that any So there is one thing like when you said it's pretty important to know what applications we want to Put in containers and that's that's one thing that We're also trying to do within this MTA program like modernized traditional applications And this is more for people who have lots of applications. So For for many of us we have maybe just like one big website or maybe a dozen of applications So that's that's not us, but for the organizations that have hundreds or thousands of applications Finding out which ones are good candidates to be put in containers can be just by itself a daunting task So we we worked on tools for that to can do some kind of audit on on this Application landscape so to speak and say okay with a bunch of pretty straightforward heuristics We can kind of score these applications and then you can Start working on the ones that are on top. It won't tell you exactly. Yes. No, you can containerize that or not That's not exactly the point. It's more to know immediately. Okay. I'm going to Get a good overall and leverage benefits and etc etc on this something on this pool of applications So I can start working there and then walk my way slowly To the applications that will be less good candidates, but eventually we'll get there as well so so in European there is no Restriction to put any workload inside a container environment You don't you don't feel that there is room anymore for virtual machines or I think there is no restriction But it doesn't mean there is no room for virtual machine The container has to run somewhere anyway. So sometimes it will be in a VM. Sometimes it will be on bare metal Sometimes it at some point it might be in some abstraction that is even different so I mean it for now it's it's still a kind of In the sidelines, but For training purposes, we have a lot of things that run darker in darker So we have containers that run on top of containers and I think those containers then run on top of VMs That if you run them on Google, then they will run in containers So that doesn't mean that we should encapsulate thing like all the way down But I think at the end of the day everything can be in containers There is no doubt about it. Now, should it be that the answer kind of depends on how you want to look at things Like if you have a hammer, it's better if everything is a nail So if you want to manage a large number of containers, it's great if you pick something that is very well suited for that and The last few things that don't look like quite containers, then we will make them look like containers So that but but it doesn't mean that you have to put everything in containers Yeah, and one just one quick comment We have noticed that some customers prefer to put containers in VMs for Isolation and for security reasons they have already existing security policies in place that it will take a while to change them because usually comes with Regulation changes and we all know how long this discussion can take and for the time being because they want to embrace containers for Their development of their application they prefer to isolate them within VMs and fulfill all their regulations Just as a comment. That's that sometimes is mandatory So what one thing and Simone I actually want to kind of ask you a question a little bit here as well because it's interesting To talk about the difference between CAS and pass and I think you kind of talked about it and correct me Please but pass or more for I guess the DevOps and and being able to manage that and CAS or more for your Your people who just are specifically working on container applications or let's let's have a little bit of a talk about that because I think this is what really interests me is the groups of people who are going to use these different layers Right, so as we start to build layers on top of the I ask It is about trying to bring together your DevOps and your app dev and your usability experts and your users and your Product managers and your communications people and all the rest of it So I'm interested to see if you're seeing anything from your customers in regards to those kind of different Modalities right so the way we implemented and that's probably specific to suicide. It's our cas solution It's relatively quite simple. So you have an operating system the container engine and our orchestration for that everything on top of this It's something where we work with partners and the custom then customer can choose between the variety of partners that are already out There while the past solution comes already with CICD included with log management with monitoring already included Whereas a customer you can not necessarily Choose any other solution. Excellent. We've got some we've got some audience questions. Yes I'm finally engaged you everyone woke up. I Just want to ask the panel about how do you think about complexity in the in the With the introduction of containers and the virtual machines and so on because obviously complexity increases with With having several layers, but also it kind of seems it decreases in some sense portability is obviously a Keyword here, but in my ideal world It would be better if we didn't have to to do all that portability stuff if we could make the applications portable And have just one OS and everything would run like that But yeah, that's not you ever seen the xkcd cartoon where it's like, oh, you know all we need is another standard So there's 12 standards and now there's 13 standards. So we're already into that with containers We had that with VMs and we have all of our different, you know Distros, but I think one of the to address the complexity question. So this is my two pence The the thing that we really right now I feel like we are in a phase where we look at VMs and we say can this VM go into this container and That's a very one-to-one pattern ratio And that's not the way you want to actually build serverless applications or 12 factor applications And we're in a phase now where? Everybody's starting to have this debate and just conversation around what are the additional features you need built around the container You know whether it's it Whether it's you know a heavyweight orchestration system like Kubernetes or something a lot more lightweight like nomad or swarm Whether or not you want to be able to do the stuff like Netflix is doing with CI CD You know so you have 20 applications out to all your different sets of users so you can be collecting data I think that's the place where over the next five years We've got to figure out what do serverless applications look like and all the rest of that and that's a much more That is where it gets complex, but just in terms of you know customers moving over into containers I don't think that's too big of a deal then there's a whole other level Above the pass and cast and this this is kind of why I'm asking this question right now It's because I don't know what are the things we're gonna build on top And I'm not I'm not sure if you guys have seen anything or any use cases yet or So before we move to another question I do have an answer What we have seen so far is that the complexity is already there and it started from developers trying to push into getting root access to the infrastructure and They could usually they don't get that and then they found lots of workarounds for that So what we are trying to talk about here to it you it's actually Standards that that will help the CIS admin and organization the infrastructure layer to put some rules in place And still let the developers have the freedom of innovation that they are looking for Complexity it's already there in my view. I'm not sure we will have enough time to take all the questions But the panelists will stay after sure So for me the specific question about Caz versus Paz Paz when I think of all the passes that came out I mean first of all cloud is a nis standard and nis define what infrastructure services and what platform is the services and When all the different people who created different solutions around those credit them They were following those standards Containers as a service kind of came about and said well We're going to ignore all that stuff But it doesn't have feature parity as a result to infrastructure as a service and platform as a service for things like self-service for example their metered service, so That's kind of what I wanted to address from the panel is like how do you how do you go from infrastructure as a service to something like container as a service when most containers of service does don't have all of like the basic Features of infrastructure as a service because it doesn't follow a standard In our case you pile them up So we have our container as a service solution running on top of an Infrastructure as a service and then you get all the benefits and all the goodies from you know storage network management on that layer And then in our case we use Magnum and put Kubernetes and build it on top And one of the things I think is really important to draw out here is again that who uses it, right? So in that is layer you got sys admins network engineers dev ops You know and all the rest of it up in the cast and pass layer you're seeing those usability app dev Community manager people and stuff like that. So one of the things I think we're really trying to do is actually break down all the barriers You know so actually trying to get your application developers to talk to your dev ops So that they can understand each other on each side of their their layered fence because right now that I think that's where a Lot of organizational break down is occurring is that they don't actually know how to how to participate with one another on that So more more concrete questions or Kubernetes or swarm both both and nomad and Do you think there are messos? Yes, thank you Do you think I'm sorry, let me just do you think there are comparable products or they're like things of their own and we cannot really say Formulate the question like this one versus another. I think that they're comparable just like EVM and Xen are comparable They they have differences. They have strengths. They have weaknesses. I believe both swarm and Kubernetes communities are kind of looking at each other and converging to this central position where They they have the upsides of both So my answer would be both like they depending on what you need to do or yeah I think also what you want to do is look at who their customers are going to be so messos because it is got that apache Hadoop background they've got a bunch of customers are doing a lot of big data stuff, right? So that's what's going to happen is that people are going to come around these things and their use cases are actually going to Form the product and we need that because we we need more diversity in this space We're not we're not having less and less things. We're having more and more use cases. That's the biggest problem Yeah, so we need these different Cases slash passes to be able to have vertical markets inside of this space and understand what those customers need inside those vertical markets So getting again all of these what's really cool is every single one of them has a community You know, we are finally in the era of community where go and hang out with that community and see who's there And if you know birds of a feather, you know flock together, so that's the place to go David you said, you know, it's not about either or it's not IES or VM Ies or containers, but it's it's actually and you know, I think that's reasonable But you know if that were to happen you really need to have a single Orchestration layer which can actually orchestrate, you know that kind of mixed payload environment right now You know, there's one of the challenges we're dealing in our company is you know We have cuban it is we have open stack if you have some payloads running on open stack some running on cuban it is You end up building another higher-level Orchestration layer which would talk to cuban it is an open stack and maybe another one later on. How do you think? That evolution would happen would open stack end up You know orchestrating containers as well or would cuban it is also support VMs in future Look, I think all of those are a possibility right now I don't know which way it'll do to go Yeah, no orchestration absolutely essential and you know you look at Nova and what they're doing They have such rich history in the way that they have managed You know VMs and bare metal and all the rest of it the way they're ingrained in containers But at the same time there are very definitive use cases for you know More heavyweight solution like kubernetes to be able to do the full-blown pipeline stuff versus just more simple Orchestration tools like heat like salt like ansible so I don't think we're gonna Converge in on a single you know It's not actually gonna be the board Borg is an interesting work that kubernetes came from Borg because we're not gonna have a single orchestration language People have their favorite orchestration and configuration languages and you're not gonna give those up Just like you're not gonna give up your favorite programming language, right? You know, it's it's something you have in your your arsenal of tools. I think we're out of time Okay, thank you very much for panelists