 Good morning. All right, so let's get this panel going So thank you all for coming Guess we're running a little late. We have around 40 minutes trying to make the most of our time My name is Ashish Nadkarni. I'm part of IDC's research team. I'm based in Boston And it's my honor and privilege to be moderating this panel. We have an exciting bunch of folks here and hopefully some questions to get the Conversation going I Want to make this interactive? It's still the morning some of the some of you guys have probably not caffeinated yourselves enough But I want to keep this interactive. I have a bunch of questions But really, you know, I want you guys to ask this is your time to ask questions as well So let's see how it goes. I'll walk around and hand the mic So if you have a question you need to ask just raise your hand and I'll walk over to you All right, so with that said a couple of housekeeping items I am told that there are t-shirts to be given off after after this panel So feel free to grab a t-shirt on your way out another thing to note is Skeletee is going to have a demo Not the session afterwards, but the one after that so feel free to come back and See how skill it is doing the demo. So all right, so with that I'm going to hand it to the panelists to introduce themselves and let's get it going. Thanks. So Steve you want to go first? Thank you. She's good morning, everybody. I'm losing my voice. I apologize my name's Steve mule and the executive director of the Elastic cloud infrastructure at Comcast. We run OpenStack in production at scale I'm very happy to be here talk about the past and future OpenStack. Thank you Mark shuttle with from Ubuntu and canonical Hi, I'm Jerome the cat this year of Skeletee. Hey, I'm Matt Curley. I'm from HPE and I work with big Scaleout specific servers for things like object storage HPC and big data Awesome. So the team for today's conversation is about OpenStack at six years and what has worked what has not worked so it's a conversation around what You guys are seeing what we are seeing what the panelists are seeing on the six year anniversary You know, we got news from the user survey that 65% of OpenStack deployments are in production now So clearly something is really working And so the question I think I'm going to start with Steve being you know a user of OpenStack What would you say has worked and what has not so maybe you can sort of give a user perspective and then we can go to The supply get the supplier perspective Thank you. As I said, we run OpenStack in production at Comcast. We've been running since Essex so four years or so We have a very widely diverse multi-tenant set of customers. So That works very well. The core of OpenStack is extremely robust It works. We've been very happy with that I think the challenges we've seen what doesn't work so well is Scalability, we're really kind of pushing the limits there and and the limits are not that high, you know, certainly hundreds of physical servers Everyone we talk to all the involvement in the large deployment community Everyone is hitting the same pain points around scalability and I think just general maturity of the the OpenStack development process before I took on This team I was a customer of our OpenStack deployment and just silly little API compatibility or tool compatibility Things that just create additional work unnecessarily. So I think attention to detail in terms of Backwards compatibility Upgradability Those kinds of things really maybe haven't been a focus of the community And I think that's something where potentially vendors can help drive that by making the emphasis This is how we become an enterprise grade Truly usable platform. We've invested a lot to get to where we are We've been very happy with the core of OpenStack But I think there's a lot of work to do going forward to build it out into sort of the the tool that it could be So I'll take the next part of this So my perspective is a little different from you know you've seen some of the sponsor sessions etc from HPE around here and They're certainly working to sort of bring some of those robust vendor pieces into it the conversations that I have You certainly have the kind of scale out issues that you're talking about and really it's been kind of like two classes of customer They're ready for it What we work with on the solution side with things like scalability is there are people that are absolutely ready for it And they're willing to put the engineering in and build that expertise But there's been a real chasm with our conversations with people that say hey, I just can't figure out what's out there I can't sort of filter the noise from what's really going to work for me and Some of that product stuff is just not ready for that class of enterprise customer They want something that's very customizable, but they're a little bit lost still as to what's out there Maybe to offer another vendor perspective And actually we're both a vendor and a user because we have a fairly Larger test and a continuous delivery process. That's actually running on open stack to Two thoughts first of all as a vendor It's actually not easy to understand where to put our efforts to complement what you described So we realize that open stack has a lot to be done a lot will be done by the open source community And that's great and a lot can be done by the commercial vendors but where we're Struggling finding where does the community want us to work for and we would focus on that if we knew Mark So I think I can support the view that the core of open stack has worked well I think it you know you ask someone out there in our industry on the street What's open stack that sets an infrastructure as a service and in that spirit? It's it's cracking on really well I think what has failed is the broader big tent and I think there's a reason and To be clear, you know one can love a project and still offer critique. I think it's important And I think the reason it it's failing us is two-fold first Much of what is being put into the big tent has nothing to do with the mission of infrastructure as service, right? And so that makes for a confusing message to users It's branded open stack and they think open stacks and infrastructure as a service and this stuff is branded open stack So how does that work? I think that speaks to the your point about where do we participate? Where do we contribute? I think that's the first issue in Linux. We have a very clean line You know, there's the kernel and Linus is really good at saying that doesn't belong in the kernel go somewhere else And I don't care where you go, but go somewhere else, right? We don't have that Crisp view here. I think that's a failure and and we should address it The second reason I think the big tent is failing us is because the reality is the motivations for many of those projects are vendor motivations and not user motivations and We should be honest about that, right? My view is that the vast majority of those things will fail call it the collapsing of the big tent And the headlines then will be confusing the headlines will be that this is all about the failure of open stack because we've called those things open stack Right, but the reality is open stack is going to power on right in the same way that the internet didn't stop in 2000 right Dot com nonsense stopped, but the internet kept powering on and I think that moment the sooner it comes the better because you know That is essentially going to make things much clearer for users as to what is open stack? What is important about open stack and that that is successful? Faces challenges around scaling and so on but we can deal with those challenges, right? What we can't deal with is the endless addition of complexity for no real reason that essentially tarnishes the image of the core So so mark what is open stack or what what should open stack be in your mind? Yeah, that's a great question Right to me It's super straightforward and I couldn't believe that it took us two years to sort of even talk about the topic, right? but open stack is virtual networks disks and compute right with the critical supporting infrastructure and that I would say is is identity security and And I would put things like Barbican and designate into that because they're part of that critical infrastructure Outside of that you're in user space, right? And you really should be somewhere else the acid test would be to say Can you run your stuff this other stuff database as a service messaging as a service blah blah blah bullshit as a service Can you run that on AWS? Right because the only things that are going to matter to a CIO Right in that foobar as a service category are things that they can consume on in a hybrid way right on AWS on Azure on Google and on OpenStack And I think we're missing the the DNA to draw that line really clearly and firmly So that's a good question. So I just wanted before we jump back into the questions How many people here have open start the open stack deployed? Oh It's a good chunk. So do you guys agree with what what is being said about the challenges with open stack? Yeah, awesome. Thank you So so then I guess I want to throw it back to you and say you see that there's a one of the complaints that We hear often is the quality and consistency of projects is Not the same across the board and I'm sort of wondering, you know So if you take the storage projects you take some of the other, you know Networking projects. They are at different levels of maturity different levels of consistency Quality and that kind of you know impacts the users from a deploying from deploying the whole thing, you know So how do we why is that and how do we take care of it? So who wants to go? I mean Matt, do you want to I'll start a little bit on that So, yeah, I mean, I think when we talk about things like guidance, there's too much vendor BS coming into it It just collapses under like, you know, some sort of cash priority if there's no real sort of guidance Then you end up with a bunch of people saying, okay, I want to throw everything the kitchen sink into it So I think we've seen a lot of this sort of stuff where we say yeah I mean bullshit as a service or people putting things together and We don't have this sort of come back and say, okay, what do we want to do with this? What's the overall vision? I mean, I actually have an overall question I mean from the community standpoint, I mean you talk about things like that in the kind of the core Which is very much what I see and what my customers want to say How can I get that stable picture but for some of these things? I mean, there's a bunch of people out there that believe that it's an overall integration engine Which I think is a very different sort of view and I'm not sure how we get those people together and aligned I think that's part of it also with all these different projects. It's kind of hard I mean you need real leaders kind of driving this stuff So so I think a good way to look at this and I think Mark really you hit it a good point Which is the whole kernel user space and I think a good analogy sort of take that one step further is The kernel versus the Linux distro right to some to most consumers Linux is a distribution and Ubuntu is very successful and so on Open stack is In some sense the core of open stack is the kernel But what users want to consume does not have to be defined as open stack. I think getting to your point And so I'll just talk about how we do this in Comcast before I came onto this team I worked on a product building databases of service But we understood that that was a customer of open stack and there was no desire or intention to move that into The infrastructure team right we provide it as a multi-tenant service to our customers And so our customers see an environment This is kind of like AWS where there's the value add of all the stuff dynamo db and route 53 and so on above The infrastructure, but you don't have to brand that as open stack And I think it's a temptation to say let's and I don't I'm gonna you know It's not about Sahara or Trove or whatever Specifically, but I do think it's very tempting to just add stuff in because of various reasons that don't serve the goal of Focusing that core our open stack deployment is as core as you can get it is literally just the core stuff We don't deploy any of that extra thing We're looking at some of the Barbican to designate and so on but it's literally just the core and it works great And that has made our life so much simpler than getting into people come and ask us They say how about this shared file system stuff and I'm kind of like No, you know, you know, we're not gonna make that part of our core infrastructure, right? we provide key primitives and We will have a higher level environment, but that is not open stack So a couple of years ago. There was this whole deaf core initiative and I don't think they talk about deaf core much here From your perspective, you know, the I so I agree with you Steve at the core and and what mark you mentioned How do you how do you implement it from a vendor's perspective? What what do you say as the way to go about? Taking care of the core versus sort of looking at it and sort of layers of an onion or concentric circles However, you want to call it me do if it's for me as a vendor. I'm very confused. I Said it earlier. I I wonder where where the community wants us to be and where you know where they want us to develop Swift is a great example is a swift an API which I think that totally belongs to open stack or a swift an implementation So we've defined we basically plugged our own storage implementation behind the open source API, which we think is the right model But I know that now everyone thinks so and You know along these lines I fully agree with Mark said earlier I think that what users want and this is especially true for large enterprise is the freedom to be able to deploy their application on any cloud Whether it's an open stack cloud and AWS cloud and Asia cloud or whatever Because some other flavors are going to come out and I think that open stack has been Running at as its own island and in some ways it was the first one. It's the first one that put an intention out But now it's not the only one anymore on the planet and and there's a real question of okay What are the interoperability? You know and just to keep in storage because this is the field that I know the best We we think that the Amazon S3 API has become the de facto standard That that's what most people want and yes There are people who started deploying open stacks So they won't swift just because they're used to it But if you look at enterprise who are starting from scratch today, and that's the vast majority They're natural tendencies to go to the S3 API. Sorry So I think this question is only confusing if If you're talking to people who have a vendor perspective, right? Look at the Linux kernel we have a VFS layer and it's an API and If you you can plug in lots of file systems Because they fit into the VFS layer and as a file system vendor, you know You're not going to get into the kernel if you don't fit in with the VFS layer, right? And as a user, you know you can add all these file systems because it'll fit into the kernel This is only confusing because of vendor politics and dynamics and that's only a problem because leadership In open stack is a political exercise Now if you just want to think for a while, you know, this is a choice. We all made but it's a choice. We can change right People love comparing Linux and open stack and I'd like to ask how many people have participated in the election process for a Linux kernel subsystem maintainer None right So this is a choice that we've made We haven't we have an excellent foundation Seems like seems like they could employ subsystem maintainers Who could be vendor neutral and at that point you'd have no problem getting an answer to your question, right? Yeah, I welcome that day So I want to kind of do a quick poll. Oh, there's a question. Okay. One second I would like to ask the panel is open stack assistant assistant administration's toy or something if you think that nothing from Application or any community is going to impact open stack So I think the statement that BigTent is a disaster is I would argue that no Like we have opens OP NFE open platform from NFE We need something from neutron for the SDN controllers to be able to operate interoperative Open stack. So how can you say that BigTent is a disaster? So Not only that the additional thing is product working group is definitely working towards different verticals to be operated on top of IS whatever you call it but that means there is a requirement from the industry To open stack to grow beyond the six years what it has done. So I would not agree with you on that that We need to have Testing critical testing of open stack through verticals So, thank you. I think you may ask some good questions I think OP NFV is a good example of Something that we absolutely do not want to bring into open stack And so yes neutron has to provide APIs But neutron has APIs and if the APIs are missing something then we'll add that should be added to the core of open stack But going back to your initial point, I don't think there's a any problem here You know as Mark said open stack is a core set of systems components that work together to manage your infrastructure That's all it is. That's exactly how we position it at Comcast. It's the infrastructure orchestration for all physical compute storage and networking and That doesn't mean that we don't care about application Developers our middle word. I know it's not a trendy term, but everyone knows what I mean colloquially Developers of course we do but that's not the concern of the open stack community or the open stack developers their job Just like Linus doesn't care about people building JVMs or people building Apache or Firefox or whatever not should he the open stack community needs to be able to focus on building that core infrastructure orchestration service or environment and Yes, there are Working groups and so on that can focus on verticals and testing absolutely critical and interoperability with open daylight or other controllers Absolutely important to but that doesn't mean that that needs to be part of open stack per se open stack needs to have that same level of Focus as Linus has that isn't to say that the way he does it should be replicated I think nobody would argue that but that model has worked incredibly well Maybe mark is you know the benevolent dictator. Maybe the benevolent dictator for life thing is I don't know Maybe it's Mark I'd be okay with that you sound a reasonable guy but I think The core works well, we should focus on that and not get distracted by political or whatever Objectives to expand and increase scope and that's exactly why we have APIs Boundaries for interconnection to other things and they can stay in their own domains So let's bring me to the next set of questions on influence and meritocracy So we have a bunch of open stack users here. So I like a quick show of hands on You're here. You're at the summit Do you believe you have an adequate level of influence over what's happening at the open stack? So who believes that there is influence on your deployment? I mean you have so wow So so you do believe that there is more you could do do you believe that you can do more on Sort of influencing open stack in a direction that you're taking into account your you know your feedback anyone Sheesh, can I ask a quick sort of follow-up question? Yeah, just because of the show of hands was so small How many people contribute code to open stack? Right, so that to me that is influence and I think that's really if you want to influence contribute code I mean, it's open source. It's open. So I mean do you do you feel yet that you have a way to Contribute code if you would like to contribute code Yeah So so as a technical guy I could have more influence by spending closer to a hundred percent of my time doing that But I don't think that's the kind of influence that you guys are talking about Because you're talking about a kind of influence which would be more than a hundred percent of your own personal time How many folks would like to have an open stack where without hiring anybody who knows anything about open stack You can just say there's a bunch of racks. Please make me a cloud Right so that my developers can be productive. Well, hallelujah. Let's work towards that And in that process, wouldn't you like to be able to choose? Oh, I just want to use scale at ease back in storage for object storage Or I want to use Swift stacks or I want to use another vendors. You see Oracle doesn't matter, right? Make make make technical choices don't have to Get into the weeds about how all of those things fit together and get your developers focused I think on what really matters which is the apps that are critical for a particular operation, right? Particular mission who would go with that as a mission? Yes, okay So Talking about meritocracy and I think the foundation likes to talk about meritocracy in open stack and you know the ability to kind of You know what gets elected, you know in sort of the self-election process So what you know would you say is how is that process working or not working from? Code contribution perspective, you know, I believe all three of you in some way shape-of-form Or your teams contribute code to open stack. Do you believe that that process is working not working? I'm certainly less experienced than some of these guys, but I It's available there. I think one of the hard things when you talk about there are very few people who are actually Available to say I can spend a hundred percent of my time there I think there is however some community contribution you see in other areas That also has been feeding in I mean people are giving real feedback about their deployments I mean a lot of sessions like that are very popular here and that is a real contribution to the community There's documentation stuff honestly a lot of the API stuff the code may even be good But without some sort of clear information on how you can extend this stuff. That's confusing a lot of stuff in vendors So I think there are certainly big missing pieces But from my perspective as a vendor who can hire people there It's not bad to contribute in now some of the projects It's a little bit harder to actually say hey, can we really treat this in a bureaucratic way and get features in? so we've been contributing code around the storage projects and I mean the feedback from my team is that that process is working very well I don't think we have a problem there I think we have more of a problem in and this has been the theme over the past 20 minutes What project should be in and what are the project but better defining? So it's really a leadership in terms of direction Rather than accepting and discussing code contribution to my knowledge. This is working very well We should think a little bit about leadership and what leadership is about and tough decisions So I have a view which is that nothing important can get done without taking tough decisions What makes a decision tough? It's not that it's technically difficult It's that different people want different things Right, so if you believe that and I think this is the core truth about leadership and decisioning, right? Then you have to ask How can we have a process? whereby to stay a leader You must be elected Right, what are we doing? We're saying that the best way to stay a leader is not to make decisions, which are tough ones, right? This is what we've built ourselves right The truth is people are afraid of leadership Because they realize that every now and then a decision will be taken that they don't like or don't agree The flip side to that though is that decisions do get taken It's just that it's unclear who took them or why you know you get that old Well, we took that decision as a community. Well, I know how that ends, which is badly, right? Again, look at look at the Linux project Linus is able to take decisions. Sometimes you get him's right. Sometimes you get him's gets them wrong, right? But it's very clear that a decision will be taken and That's what I think we need to think hardest about here. So I have a quick follow-up to that question What happens when Linus is no longer in control of the whole Linux? project who takes his place does it then That's a succession question. Yes, that's a succession question, right? And I'm not to derail this More panel so I want to talk about not talk about Linux, but I Is is Linux going to suffer the same fate as open stack is if there is no if it becomes a committee? Yes Right. Anyway to be clear first, I'd say open stack is the core is working well There's a lot of obfuscation around that that confuses the core is working. Well, I think we all agree, right? And what we're talking about here is how we can go even better Right, and I would say the key thing to think about is how do we take tough decisions? Like how do we give people a mandate to take tough tough decisions and it's not by voting every six months So so then this is leading into the next set of questions on decisions, right? So I would like to ask by So starting on the question on architecture So what in your opinion needs to change with open stack if it has to stay relevant in the future Steve? You want to go with from a user perspective? Yeah, I think it's the point that I made earlier is that it's really more attention to The realities of deploying open stack at scale in enterprises and service providers But just let's just call them all enterprises service providers a type of enterprise and that isn't necessarily About scale in terms of number of things or that's a big part of it. It's what's happened once you've deployed thousands of things How do you manage them? How do you monitor them? Are you gonna require all those things to be? Totally reinstalled because that's the upgrade process. That's insane. That's impractical I see this even in I mean some vendors have basically codified this and said if you want to go from community open stack to my vendor open stack the processes install it Greenfield forget your installed base of 50,000 VMs just nuked them all and I mean and that's craziness and the fact that a vendor could think that that was even a Reasonable approach and so to me it is a cultural thing that we all have to and I get that like a big part of I think part of the success of open stack has been that it would have been so easy for any individual contributor to stand up dev stack or stand up a cost of Four servers or ten servers and it works really well at that scale But they do not have the perspective of running open stack on thousands of servers across Five data centers possibly across multiple continents You know, I'm going to say the same thing, but with an economic perspective I think that we need to have a hard look at the cost of deploying open-stack at scale and We've actually done this at skillity and we're not a huge scale, but we we have a fairly large probably about 100 server deployment And it's not cheaper than alternatives Right now, I mean when you add everything and all the skills that you need to actually deploy it and make it run It's not cheaper. It should be. I mean I expected it to be before doing the analysis I think that's that's one thing and that's really we're talking about the same thing just from the technical perspective from Did your modeling? Did you talk at human? Yeah, yeah, so sorry. Can I just address that because I think it's sorry just quick response Open stack itself is cheaper. It's free Operating open stack. No, but this is a point, right? How you operate it and how much capacity you use and how right if you provision it, right? And you bring in your tenants and you can do multi-tenancy then you can get the efficiencies that make it cheaper To operate, but you're right if you just deploy open stack onto a bunch of servers because Operating that is not necessarily going to be cheaper, but I think this goes back to the question of What is open stack open stack is It doesn't cost you anything and If you have if you build expertise and you build the team then it can be very cost-effective and we found that it absolutely is cheaper I'm sorry when I look at the cost of something. I'm not interested in just the cost of a component of it I look at the cost of running an infrastructure I understand TCO and we've done that and it is cheaper for us if we have the right level of Utilization and provisioning and so on so Steve am I hearing you correctly in that it's really the scale and the economies of scales That's exactly that's exactly it. And this is why scale is important and maturity to make it Cost-effective and I will say the single biggest we have the different perspectives here Somebody who is paid to operate open stack and somebody who pays people to operate open stack The economics are the issue and the human economics to your point are the issue When you said open stack is free, you're exactly right We're in an era where scarcity has shifted from software To operations Right, think about it software used to be incredibly expensive Now it's free Operations used to be cheap Now they're incredibly expensive and that is something very few institutions have understood This is why we have this relentless focus on the encapsulation of operations knowledge in charms as code So that we can up level instead of just collaborating around The applications which are not the bottleneck anymore. We collaborate around the operations I was very serious when I said an organization should be able to point at A bunch of racks and say give me a cloud and have zero people in the building Who know anything about Nova? Swift neutron Cinder That's not to say that the best organizations won't employ those people because we need to advance the state of the art at the scale of Thousands those charms should encapsulate the knowledge that's being achieved at the cutting edge But that operational knowledge needs then to become consumable at your scale free of charge That's the magic that people haven't yet started to really think about So yeah, no, yeah, I'll add on to that I mean I think what I hear in a large sense here and what we hear from our customers is not that open stack itself is unattractive But there's a life cycle management problem that they haven't solved That doesn't necessarily need to be part of open stack and some of this very custom per the customer going in there but I mean The bare metal piece and where they're trying to tackle stuff like ironic getting the charms and stuff in place Making sure that everything at large scale can properly be tracked probably set up. That's big issue But let's let's just go a bit deeper right here. We're all gazing at our navel about open stack Open stack is just one example We're in an era now where every interesting application or piece of software or domain is What we call big software, right? It's many pieces of software and those pieces change every six months and that's true of Apache Hadoop It's true of Kubernetes. It's true of Mesosphere DC OS. It's true of machine learning It's true of all of the container application Orchestration systems. It's gonna be true of every interesting thing the CIO wants to get done over the next 20 years, right? Big software, but we're still attacking that with a view that you need to hire a bunch of people and create a bunch of automation Right, we have to we have to step back and realize that all of these classes of software are really taking us into a domain where operations are the challenge not access to the code So I can I do yeah, good. I just wanted to actually get into the This I think this is the most important point and this is why I think maturity is critical as the focus for the Open stack community and scale When I look at what I am paid to run open stack And I love doing it And what I will say is when I say open stack is is free because when I look at what my operations team does I have a team of about 10 people do operations Maintaining open stack is not what they spend their time on and so part of me saying open stack is free is Is a relative comparison because the things they spend their time on they would have to do in other environments, it's hardware maintenance, it's Tenant onboarding. It's a lot of stuff that is not specific to open stack And that's a test to the quality of the core of open stack The other aspect of this and this was exactly what you hinted at is it is the economies of scale and it's the utilization We get value by making sure that we run at a high enough utilization on our very expensive Servers to justify the cost. It's very easy to get sucked into Overpaying for capacity of physical infrastructure that you just don't need and that is when it absolutely will not be cost effective, so I think What you said mark is absolutely correct we have to take our learnings and Feedback and this goes just go back to one of your questions about Maritocracy and contributions math you made a really good point contribution does not just mean code in the purest sense We have a huge Community of people on our team that have done a lot of contributions to documentation in Operator best practices and it's nice to see the community recognizing them as You don't have to be a developer to get contributions We've had people elected to the user committee recognizing their contributions in other domains than code, so I Just wanted to give a little bit of insight because I think it addresses many of the questions is just understanding Actually, how a large enterprise uses open stack and why I can sit up here and confidently say to me Open stack has been a very successful Product when I look at the core as a tool for building infrastructure for a very large number of production Applications. Thank you. So I'd like to open it up to questions from the audience Wants to go gentlemen. Thank you for taking my question My employer pays me to work on open stack to do to manage it lift the mic up My employer pays me to manage open stack at home. I use it as a means to an end It's products that I want their things that I want so as a user I mean I use it at work, but you know, I'm interested in more than what I can do with it as a means to an end What would it take to influence that to be able to say? Okay, well, I want I want my desktop in the cloud somewhere And I want to be able to access it from anywhere those kind of things as just a single contributor How can I influence that direction the direction of open stack in that way? I would say and this may sound Maybe not the answer you want to have but maybe opens that isn't the right place to contribute in that direction I think what you describe Maybe goes back to the point about what's core and what's not and I think many of those things You will have a lot more success And I say this because I think it's a positive answer if you want to do something that you don't see that you see is is way out of scope The route to success for you is not to try to get it to be in scope. It's not to shift the scope of open stack It's to go off and build your own thing And hey, maybe you are onto something that no one else has thought of and you can be very successful Because you don't have to answer to committees and elections and all that kind of crap And I think that that's how I would approach it if I were you For your specific case running your desktop in the cloud I'd say apps open stack absolutely should do that that that is just a machine running in the cloud I Think in terms of you as a guy that who's paid to understand how open stack behaves operationally what I would like to see is that you are able to contribute Operations code that makes Andrew's life easier that makes Luke Jerome's life easier Because what you're learning is not going into documentation that someone has to then go and read and understand and remember In order to make their open stack better, but it's actually going into code which everybody else just runs, right? And so that's that's why I think open stack professionals are important because their learnings should then be shared as code Just like Nova is shared as code Oh, yeah, so we're time for one more So we talked about technical we talked about politics, but if you look at it from a business perspective How does that apply to what you define as what you want to be? Core which is just disk storage You know networking and virtualization From a business perspective people want database as a service and all the projects that have come in in the last two years So I think you should divorce the code quality problem from core just like Linux does it with user space programs And where do you see from a business perspective? Where do you see the number of projects topping off in this project that would be acceptable to me as part of open stack? You can't possibly just want database as a service on open stack, right? But hold on so I think at a holistic level it's a question here You you want that stuff on public and private cloud and realistically The important public clouds don't talk open stack, right? You want that on AWS? You want that Azure and you want that on Google? So I'm saying what you want is exactly right Everybody thinking that this is the forum to make it is exactly wrong, right as open stack as a community We should benchmark our success not by the fact that we have a project that promises database as a service But that we have a commitment from the best database as a service vendors in the world Who are currently on the public clouds that their stuff will run on open stack, right? Ref stack is more important than Trove, right because ref stack Allows a winning database as a service vendor that gives a CIO real options on lots of public clouds Also to make the commitment to do that on open stack, right? And so we should benchmark open stack by the number of global winning Projects as a service on the public cloud that also put open stack on their on their Sort of badges of success right on this certification list So your your requirements are exactly right But but this project is not the right place to fulfill them if we do our job, right? Then the experts out there in cloud will come to open stack So with that I think we're out of time. So I'd like to have you guys give us closing questions I was closing comments. I'm sorry. So Matt you want to go? Yeah, no, I think we're really talking from a perspective here and overall thing we're seeing this is more of an architecture as opposed to an implementation sort of story and From a business perspective, I mean I think there are a number of features that even in the core in terms of HAA multi-site and stuff to be brought in But I really see going forward that we are on the right path We can fix some of these community problems and bring this together There'd be a better sort of leadership story I love the the Miranti slide yesterday We're just at the very very tiny beginning of a big journey And I think it's very important that open stack realizes that there will be competition. There will be other cloud systems and It's very important to include a perspective of the whole world and not just to be focused on the island Yeah, I want to finish by supporting the very first comment that was made from the panel by Steve But the core of open stack is great And for any medium-sized institution today, it is sufficient. It works. We are at the end of the beginning Right and nothing that anybody said here is Questioning the relevance or the purpose or the mission of the core and Steve is exactly right to open with the statement that open stack works, right everything we've said here is really about how to get better and questioning some of the blah blah is a service around it Yeah, I think that's right. It's easy to lose sight of the You know the core the point the the central tenant, which is yes, it works. It's good and we should all be proud of that I think the other thing I'd say is Mark is absolutely right that Your goal as customers as service providers to your enterprises should be running applications that are best in class applications and those don't have to be part of open stack to Add value to your customers And I think the thing that we didn't really talk much about is what is the customer focus? I think that's an important part of being an enterprise cloud provider is we have to all start thinking Customer first right what do our customers actually need of what we're delivering But that doesn't mean it has to be filtered in back to open stack blindly and said we need to make this part of open stack Thank you and with that I'd like to thank you all for coming Thank you. Thank you to our panelists. Thank you