 Okay. You can all hear me. So, thanks everyone. Welcome to this panel. And the title of this panel is Behind the Stay of the Stack. And this is actually kind of leveraging a talk that Randy's bias has been giving at past summits. And we'll be giving a version of for this summit coming up later in the week. Wednesday at 5. It's just before 5, I think. 4.50. I'm looking it up right now. Okay. So, Randy, you can talk a little bit more about that later. So, this is really a kind of playing off of that and talking about based on the things he's found about what the state of OpenStack is today. Branching that off of some questions that we have for this group of panelists to talk about where they see OpenStack today and what we can do together as a community make OpenStack a better solution. I want a better project. So, a couple of housekeeping things. 5.20pm. 5.20pm Wednesday. So, make sure you go to that. So, a couple of housekeeping things. If you have an empty seat next to you, could you raise your hand? Okay. If you're not from EMC, grab a seat. You have a seat. It might be best, Brian. Okay. Right. Well, if you're from EMC, you can fill in the seats now. I guess that sounds like good enough room. Second thing is, I encourage you to stay for the whole thing. Obviously, not primarily because you're going to hear some great things from our panelists, but also we are raffling off a service pro 3. Is it 3 now? So, we'll be giving that away at the end of the panel discussion. I think that's the first usable one too, right? So, what I want to do is I want to go ahead and let the panel introduce themselves. So, to start, my name is Ken Hoy. I'll be monitoring the session and I'm with EMC's Cloud Solutions Team. I'm Randy Bias. Hi, my name is Mohammed Khaled. I work with the Cloud Solutions Team at EMC. I'm Brian Gallagher, responsible for Cloud Management at EMC. I'm Jeff Olson, part of our Global Services Cloud Portfolio Group, responsible for creating open-stack related professional services. My name is Raghavan Srinivas. I go by Rags. I work in a small group in EMC called EMC Code and the idea is to work with open-source developers within the company, outside the company. I've been doing application development for a long time. Thanks everyone. So, what I want to do is I have some questions I'm going to give to the panel that they can discuss among themselves and then after that there will be some time. There's a microphone as you can see in the aisle over here to my left and we'll have some time for audience questions and then we'll wrap up with the grappling surface. So, to begin with, obviously, since we are at the open-stack summit, the first question I wanted to throw out to the panelists is, to date with this killer release, what do you see has been the greatest improvements of open stack from Juno to Kilo? Who's going to take that? It's all going to be you, Randy. It's all going to be me. So, I don't pay that level of attention. I mean, I'm just being honest. I don't have time to go install open-stack or manage it. I mean, those days are very far behind me. The last time I dealt with open-stack hands-on, I was deploying Diablo for AT&T in summer of 2011. Okay, so how about, just in terms of the history of the project. So, what I've seen, I think it's the usability. I think some of the great additions has been the Murano project by adding that to self-service catalog feature. It really is going towards allowing not just developers, not just very advanced technical people to be able to manage open-stack and run open-stack but your everyday users. So, I think that's something that is almost a necessary evolution of the way that the cloud platform and also other enterprise systems are going. So, I'm really excited about that, that being added to the open-stack core project and I'd like to see what that's going in the future. I'd also add, Ironic, I think that's been a good addition as far as programs concerned. Dealing with hardware substrates and making it a lot easier to deploy. I think that's where a lot of the improvements will come in terms of automation, reducing variability, making things much, much more stable. Yeah. Can I add, Kim? I hate to be on the panel and just say Ditto, but I think I'm a big fan of Murano as well. I think I was at the summit last year in Paris and some of the discussion was that infrastructure as a service is kind of boring to developers and I think Murano has made it a little bit more boring in the sense that it will just work. So, I think Murano is cool and it makes it application development a lot easier. That's one of the big things about Kilo, in my opinion anyway, that I think is going to make a difference. Yeah, just to add actually, as I've been following open-stack for the past four releases, you can see it's more operationally focused. You're looking more at the operator, you're looking more at the user, things like triple low, things like over slow, things like rally to test the templates. Those kind of things definitely help IT operators and basically that's kind of classic improvement I've seen. It's becoming more operationally focused too. So, we're on the 11th release now, right, of open-stack. I think most people who have been working at it for some time would be, it's more usable than it was obviously several years ago, but there's still a lot of users who say it's really difficult to deploy and to manage open-stack on a day-to-day basis. So, what do you think are the biggest challenges that's holding back allowing people to manage and deploy an open-stack in a simple manner? So, there's two elements to this. The first is expectations, right, and if you're going to deploy like an alternative solution like cloud-stack, you're just deploying compute essentially, right. I mean, open-stack is much more ambitious. The minute you start putting compute and storage and networking and authentication and application deployment and management, all those things together, you get an exponential rise in the complexity of the overall system. So, one of the problems is just people have an expectation that they're going to go take open-stack, download it, sprinkle it in their data center, and poof, there's Amazon, except it's just not that easy, right. And then the second piece is that, and I'm going to talk about this a lot more on Wednesday, is that we are just unwilling as a community to actually, you know, kill certain projects that are causing a lot of grief. So, last week, it was a little bit late, but I did an informal survey, I got about 70 responses, and it was very obvious that there were a couple of projects, I'm not going to tell you which, you can find out on Wednesday, that people just universally hated. Why aren't those re-architected? Why aren't they rebuilt? If people hate them and they're hard to operate and they increase the complexity of the system, why aren't they just kicked to the curb? Well, the reason is, is that we like feel that every little piece of code in open-stack land is precious, which is not how well-run companies actually do things. So, we need to figure that out. So, it's expectations not set in the right way, it's complex to build and operate a private cloud, combined with, you know, us being unwilling to make hard decisions about some of the components. I think that some people don't have a real use case for leveraging open-stack. You know, they look at it as a free alternative to some of the commercial versions of virtualization out there. Put platform two apps on it, for example, or SAP on open-stack. Yeah, I think it's this new shiny bumper that people are excited about for many different reasons, but they don't really understand, well, you know, we have to kind of develop a specific use case of why we leverage open-stack. And, you know, I think that's at least some of the conversations I've been having with people is, well, it's not production-ready. And, well, why is it not production-ready? Is it not production-ready because you're trying to leverage it for platform two apps in a general purpose cloud environment? Yeah, it's bullshit. I ran Walmart.com on Black Friday on open-stack, and it frickin' worked. Right? It's production-ready. It just depends on the vendor, the deployment, and the use case. Right. Can you talk a little bit more about, when you say workloads that are appropriate for open-stack, what does that look like? Let me jump in. So you asked about what, you know, challenges or what opportunities. I've spent about six, seven months looking at the world from a different angle. I spent the prior 30-plus years in data storage, information, infrastructure, product development. And this whole notion about workload placement, I think, is critical for this audience and this discussion. And when you look at brokering of workloads, there's not a lot of information that's offered up from the infrastructure stack to the brokering services, if you will. So if you look out today, what can you get out of a cloud, out of an IaaS cloud? Transactions per hour, cost per unit of CPU hour, or cost of gigabytes per storage per month. And that's not a lot, you know, because a lot of IT departments worry about, hey, is my application workload predictable? You know, can I meet my recovery time objective? What about backup? People think, hey, you know, I stuck it in this IaaS cloud and it just, all the data's there, right? I don't have to worry about that stuff. So I spoke at the Cloud Foundry Summit last week, and one of the, the encouragements I had is let's work with the open-stack community so that we can get more of a service catalog exposed upward so that we can broker applications in a much more systematic manner. Right now people are deploying to different types of cloud, whether it be Amazon, whether it be a vSphere environment, whether it be open-stack or other, and they're finding out that not all their business objectives are being met. So can we start to surface some things up that can do a better job of workload placement, you know, to make sure that their GRC policies are met and others are met? So in my experience, there's been two major causes of failures for open-stack deployments. First is trying to put Platform 2 apps on it and not understanding that it's a different system. And then the second is going direct to the ODMs to get white box hardware and not understanding that that means that you need a hardware certification and validation team, which you probably don't have, to make sure the firmware stays stable across all the different boxes you have. And so on the Platform 2 versus Platform 3, for people who don't get that, it's like existing enterprise applications are pets, not cattle, right? They don't manage themselves, you know, if they get sick or they start to have problems, everybody's got to be on deck to go fix them, right? Everybody gets that SAP, SharePoint, blah, blah, blah, right? Well, if you go and you stick those on a Platform 3 cloud, like open-stack is generally designed to be, the problem is that suddenly you're running on a 2, 2.5, 9 cloud, like Amazon, right? And those apps are suddenly 2, 2.5, 9 apps because the infrastructure goes down on them and they can't react. If you take Platform 3 apps and you go stick them on a Platform 2 cloud, then that works initially because you can go stick a scale-up system on a scale-up system, but the problem is that the cost basis is very, very high, so it's very expensive to run those apps. So the only people I've seen that be successful have two clouds. They've got one for Platform 2 and they've got one for Platform 3 and open-stack plays best in Platform 3, even though people like to do unnatural things with it to try to shove it into Platform 2 land. One of the things I see is, and I don't want to put anyone in the spot here, but developing applications, you talk about, well, Platform 2 versus Platform 3 and what's the difference and, you know, you have Randy's pets versus cattle analogy, but also, do you have the resiliency built into your application? Is it built to fail? And sometimes people look at me like, I'm three heads, like I'm crazy, and I say, you know, worrying about high availability on the hardware side of things really is a way of the past. I mean, obviously it has certain use cases and whatnot, but, you know, really for third Platform apps, you want to build the resiliency in, and are you doing that while you're leveraging open-stack? You're just like, you know, you're crazy for saying something like that, but in actuality, that's the way that you need to build resiliency is through your code. But, Jeff, typically you've seen actually customers who actually are, like the whole shadow IT organization where developers are actually deploying in Amazon web services, which is kind of Platform 3 scenarios. But if you're going to talk to the same developers and you're trying to bring them back, talk about, because of security and governance, of Platform 3 apps, they should use the same principles. And that's the problem is it's a top-down process, right? You have to have a good enterprise cloud architect. It's just not open-stack by itself. It's also about having a good architect, having a good team around to go ahead and help build the proper cloud. So I wouldn't say just open-stack. It's more than that. It goes to the cultural issues, because inside the enterprise, typically there's a lot of fear of failure. If you go to Amazon and you fail, like you spend some money on your credit card, right? If you go inside the enterprise and you, like, ask for a whole silo of hardware, and then, you know, you fail, then it's a big problem. Everybody gets really bent out of shape. Exactly. The principles change. That's the problem, right? So, for success in Platform 3 in open-stack land, what you want to do is you want to build the cloud inside, and then you want to let people know that it's okay to fail, and you want them to be able to use a lot of resources and shut it down, and then everybody kind of wins. Exactly. Yeah, I think, Brian, if you don't mind, you know, I know that it's not just the technology that matters, but I think the platform definitely matters, in the sense that I don't think open-stack is the only solution for third-platform applications, right? Really? Yeah, as much as we would like to think so, right? You know, we still, there is something that's missing, you know, and that's what's on top of the infrastructure as a service. And I think, you know, I was having a discussion with somebody who operates a pretty big infrastructure, right? And basically, he was saying, you know, how many of you have all your source code on GitHub, you know, including, for example, your infrastructure code as well, right? You know, how many of you have everything on GitHub, for instance, right? Few? Okay. And really, the question came down to, do we have an action plan if GitHub is down for a few minutes or whatever? And it turns out, there's really nothing. So what you want to do is you want to have a platform that kind of adapts to, you know, some of these, you know, like high availability, scalability and all that, but also to be able to, you know, insulate the developer or the operator. You're talking about the platform as a service layer, right? Because your average enterprise developer is not going to become a scale-out expert overnight, so they need kind of training wheels so that they can have a framework to code to that'll get them most the way. Exactly. Yeah. So what I'm saying is there is something on top, right? You know, and I'm kind of leading to probably the next few questions. Yeah, I want to get back to the app for a minute. If you paid attention to last November at the re-invent conference and what Amazon announced, you know, make no mistake, they're going after the enterprise apps, right? They started with less critical, started with small to medium business, moved up into mid-enterprise. Everything in November of last year was all targeted towards enterprise customers, right? Faster compute. Amazon exclusive has well offering flash tiers. They have a database as a service now. They've got better security. They're moving up into the platform space with code deploy, code pipeline and code manage. They've got lambda messaging services that they've delivered, you know, for enterprise apps. That's the, when you say enterprise apps, do you mean platform two, platform? No, I'm just saying workloads that typically are, you know, running the business that are critical. Yeah, but lambda, nobody's going to run SAP of lambda, right? I mean, that's really a platform three kind of technology. Exactly. So for example, the enterprise apps, if we talk about the highly critical apps, they typically either go to SaaS models or they're basically in-house. They don't absolutely go to... Totally. So I think that, I think where you're right on is that for the dev test and QA of those platform two apps, like your SAP test cluster, put that in Amazon's like golden, right? Because certainly you can do have multiple QA environments for very cheap and you can parallelize instead of serialize your QA system. Exactly. So let's talk, I think we've kind of talked quite a bit about the challenges. Let's talk a little bit about what can we do, particularly, this is an EMC panel. What can EMC particularly do to help with deployment and management of clouds? Whether they be infrastructure of the service, of OpenStack or PaaS, whatever it may be. How can we help make cloud deployments easy and better? I think it's kind of piggybacking off what I mentioned before is really understanding, someone comes to EMC and is looking to deploy OpenStack, really understanding what their use case is. Peel that on your back, determine are they ready to take that journey? In my experience, there are some. So we try to define people's OpenStack journey or whether they're ready, undecided or motivated. And through those very different levels of their status, we have different things that we can do for them. If they're very advanced, they've had DevOps, they understand that the applications they're developing for are ready to move. It's a much easier process to help us allow us to enable them versus the ones who say, we need to go OpenStack, we're getting a lot of pressure from the top down for us to get up our existing platform or developers are coming to us and saying we need a platform with common APIs to develop these new applications. But we don't really know how to do it. And we don't want to have you come in and drop OpenStack and hand us the keys and expect us to be able to drive it. We want some sort of transformation journey. So whether it's taking a look their apps through some sort of cloud advisory or DevOps transformation, that's one of the things that we're doing right now in helping customers is really helping them understand how can they leverage this new technology around OpenStack for success, not just come in and, hey, we'll deploy in five days and have a production cloud. And they come back to us and say, thanks, I have no idea what to do with this thing. Yeah, just pick up on the DevOps side. I think that's a key area where not just EMC but the industry, we need to come together between platform as a service and infrastructure as a service, making it much more automated, much more systematic, easy to use. I think the other thing on that is just the intersection of the marketplace. We've got a marketplace around IaaS. We've got marketplace around pass. I think simplifying that, giving customers choice of what they're deploying, what types of clouds that they're deploying to. And I think automation is another key area of focus where EMC and other folks in the industry can start to help. Yeah, for the automation piece, I don't know if people have been paying attention but not only do we have a lot of new components that are software only like Scale.io and ECS and Caspian and some of the others which is, we're selling software, EMC selling software, right? But we also have virtualized versions of a lot of our stuff like the virtual VNX and there's something coming for Isilon, I think as well. You're trying to build automation pipelines and test your infrastructure and want to apply DevOps to your infrastructure which is hard. It's very different than applying to your web application. And then we're starting to lay the groundwork with all that EMC code so that you actually have the software tools you need to succeed at having a DevOps pipeline around the infrastructure which is pretty amazing. Yeah, just to add to what everyone has said, to me, it's just not about the platform, it's also about the ecosystem. So from that aspect of it, what we are also trying to do is we're trying to partner with some of the top distributors like Maranta, Red Hat, Canonical to create some reference architecture design so that it helps people to build an open stack cloud or even down to a hybrid model because at the end of the day it's all about hybrid cloud. So we're trying to build an ecosystem to help customers solve the whole supply chain, cloud supply chain problems. Okay, so, sorry, did you have anything else? Okay. So I want to, I have some more questions but I want to give the audience a chance to ask questions as well. So if you have any questions for the panel, please come over to the mic on the left hand side over here, on my left over here and ask into the microphone, that would be great. Hecklers are welcome. I'll go first. Thank you for those great insights, right? Name and company. Oh, Giri Baswa, Trilio Data and XEMC. All right. You escaped the board? You have to leave the room. All right, so we talked about apps, right? So apps going everywhere. So private versus public cloud or hybrid cloud, which class of applications are better suited for what? Any thoughts? Okay, I'll go. So like I said in 2009 or 2010, like, you know, if you want a private cloud it's got to look like Amazon. Like the other thing that you've got that's like the Amazon demand system with an API in front of it, like you can call that a cloud if you really want to. It's not cloud by my standard, right? That's virtualization 2.0 platform. And so great. Have fun with that. You know, VMware is going to help you and it's going to make it awesome. That's good. And but if you have a private cloud, if you want to be a cloud, then it's going to look like Amazon or it's going to taste like Amazon at the very least, right? You're designing a layer cake where each of the apps go on top of it and you're getting a leverage because you have a common set of services across everything instead of silos and stovepipes. And in that case, your question doesn't really make sense because you're going to take your platform three apps and you're going to put them on public and I'll put them on private, you know, you're going to put them on a platform three cloud, regardless of where it is. And really the question is where do the where do the apps go with our platform two of platform three platform two apps on the platform two cloud platform three apps on the platform three cloud. Yeah, I think the desire, you know, from CIOs is it would be great if it was all the same. You know, and I had one environment, but the reality, Randy said it earlier, P2 apps and P2 is different, completely different than P3, not just at the infrastructure layer, but also at the developer layer, right? You know, the application development process, night and day differences around the infrastructure, how it's constructed, what it can do and how applications are built, delivered and deployed. So I think, you know, it will be by, you know, obviously people don't like to have, you know, islands of isolated stuff, but I think in this transition, it will be, you know, isolated. I think you have to ask the question, does it converge in the future? You're going from hundreds and thousands of islands to like two. Yeah, that's pretty great. That's beneficial, right? Yeah, I think again, you know, if in an ideal way, you know, everything was the shared nothing architecture that Michael Stonebreaker, you know, talked about 30 years back, right? You know, then then, you know, it's very easy to design, you know, stateless apps. Brian and I were talking about stateful apps. You know, how do you design stateful apps? You know, it's going to be a little bit different from how you would define state, I mean, how you would design stateless apps, right? So the answer, you know, I hate to say is it really depends, right? But the idea is that I think we are coming back to what you were saying, you know, try to reduce those snowflakes as much as possible, but you can't completely eliminate them, you know, try to minimize them. But the idea is that, you know, if you can get to two clouds, I think that'll be, you know, a real cool thing. There are no two clouds that are absolutely. So that's why I keep mentioning if you're talking from a very IT organization, come from a very IT background, is the way the CRs do is SAS as and past, that's how it works. So highly critical apps will go SAS. So there's no way in hell you're going to do anything because hopefully the SAS providers are doing platform three kind of scenario, which they might right, they might be using OpenStack. But I as and past as Randy and they were mentioning, that's again, you as an IT organization should have a governance model saying, hey, if this application is going to get matured or you're going to refresh it or time, how are you going to move it to a platform? Are you going to rebuild or replace? Or for, as you were saying, if a platform to stick to platform to VMware works well, just leave it at that. Just determining fit for purpose, you know, it's taking a look. You might have a Microsoft environment, an OpenStack environment, a VMware environment, and then Rackspace, public cloud, Amazon. And you don't know, you know, where to build these net new apps, where to place them, you know, your VMware guys might say, well, ours is the best platform to use it on. And the OpenStack guys might be saying, well, not really if you want to scale past, you know, the limitations of, you know, the clusters and Amazon, you know, so on and so forth. It really depends. And, you know, it's kind of a cop out kind of thing to say, but it really does. And obviously there's certain parameters that you need to go by in order to determine, you know, what is really fit for purpose, you know, depends on, you know, really digging down deep into application and actually what your future state wants to be, how you're looking to scale, really measuring what type of load you want to see with that app. I think 100% of the world's credit card transactions go through mainframe still. Like, I mean, I don't know what you're expecting. Most of the security's clearing is still done on mainframes. Yeah. Across the globe. Okay, next, thank you. Between platform three hardware and platform two hardware, the idea of having five nines applications running on two nines hardware. Are you looking at me? Looking at you. And then earlier you mentioned the idea that an enterprise going to unvalidated hardware, white box, white label is probably not an option. Where is the line drawn? It's an option. You just have to realize what you're getting into. Yeah, I'd say a couple other things, not just the, you know, the propensity towards commodity, but it's what the infrastructure actually does, right, as it relates to managing resiliency and availability is typically done by the infrastructure and platform two today. Platform three, it's completely different. So I can get the two nines EMC hardware. We're working towards it. Actually, dude, seven nines. But yeah, it depends on, you know, and it's also the model of access, right. I don't think it's black and white, but, you know, platform two typically have been block and file. There's much more object access. So I'm project Caspian. It's two, two and a half nine hardware. It's Cots hardware that we'll deliver with it, but it's running software like scale.io, which is a distributed block store like Seth that is going to get, you know, four or five nines of uptime for the block store service. That's what I'm looking for is that there is some sort of discrimination between. You need uptime. I mean, I don't think it's worth using that. Yeah, just somewhere to add some stats for the audience. So we've done a lot of cuts to the market data. And by 2018, and we looked at it, platform two, platform three, or platform three, platform two. So platform three by 2018, the overall app environment, as it relates to storage, is about a $32 billion market, right? Whereas platform two is three times that in 2018. So 96 billion, right? But the growth. It's progressive. They got that far. It's going to get that far that fast. And in platform, but the growth is all in platform three, right? It's double digit. It's like mid teens, whereas platform two is about, you know, low single digit in terms of growth. And the other interesting thing is when we cut it this way instead of, you know, platform three, platform two, we cut it this way in terms of on-prem, off-prem public, or somebody asks the question about hybrid. It's about a third, a third, a third, right? And so a third being traditional enterprise IT. You know, still platform three is being delivered on traditional, you know, kind of P2 architectures, if you will. So a third of that. A third is being deployed across a hybrid cloud. And a third is being deployed into public cloud environments. And our exabyte scale customers, we don't have very many. Nobody does yet, but we're starting to see them. They're all the web scale guys, right? It's all platform three. So I'm just kind of curious. I've been around OpenStack for quite a while now. And you guys are still talking. Name and company. What's that? Name and company. John Griffith. Oh, John. I'm working OpenStack. Oh, no, we're going to get the solid fires. No, actually, nothing about storage at all, actually. But the thing I am kind of interested about is you guys are really perpetuating the, oh, you can't use OpenStack for anything except for stateless and what you guys are calling platform three. No. Okay. Platform three is not stateless. Well, no, I actually believe it was you who'd mentioned you want to run your stateless apps in OpenStack and et cetera, et cetera. And, Randy, you said, you know, oh, you wouldn't want to run SAP or something like that. Correct. Why not? Would you run virtual desktop? Yeah, I think the answer is based on time, right? You know, right now, if you look at the pass layer, what is cloud foundry do well as an example of a pass? It deploys, you know, 12 factor apps. Sure, that's great, right? So over time, that may change, right? I think there's a strong desire for that to change. I guess what I'm saying is I think it has changed and I'm just kind of curious, you know, if you guys, apparently you don't agree or maybe you do. Hey, yeah, I don't, I don't agree, John. I mean, the thing is, is like, you know, I don't know one customer that's been successful doing that. I know customers that have thrown enough bodies at the problem. Yeah, like I was just talking to one of the world's largest credit card processors and they move their P2 apps on to OpenStack and it's not going well and they're not happy about it and they're trying to figure out what their next move is. So, but for us to have this discussion, first of all, it would take way longer than we've got right now. And second of all, you know, it's going to get very technical very fast. The reality is, is that if you look at P2 applications, though, their expectations are different. They operate in the world of the network has infinite bandwidth and zero millisecond latency and infinite IOPS and infinite capacity. They have no ability to deal with, you know, failures in the underlying infrastructure and I just seen this over and over and over again. And even if you can start to wrap DevOps around some of this stuff, which is very hard because like creating an empty Oracle database takes 30 minutes, my skill takes half a second, right? Even if you can wrap DevOps around it, like the expectations of people coming into that land are always like really out of balance. They're like, where's my live migration? I need to know that when the SAP master or whatever falls down, that it like automatically cycles to the next box. And so the expectations that folks like VMware have set around the P2 apps, it's just going to, they're not going to get it when they come to OpenStack unless they do, you know, OpenStack plus VMware, which, you know, whatever, let's not go there, right? No, let's not, please. Thanks. All right. Is there any more questions? I'm going to have time for one more. No? Can you come to the middle? Sure. Or yell really loud? So Jason Wicker with Marantis, also XEMC and VMware for that matter. So you guys kind of danced around it and, you know, I've heard this a lot and especially just recently coming into the OpenStack world, that it's not ready for production. And then at the same time I heard Randy and I heard Walmart present earlier today very much so ready for production, right? Absolutely. So would you guys say that that's really not OpenStack, more that it's the company and their internal ability and their partner's ability to be ready? So the distinction I made was experience that I had talking to people where it's that comparison with Microsoft, VMware and OpenStack and out of those situations, they're trying to make an apples to apples comparison more so around general purpose. Like we want compute storage and networking for all different types of workloads. Where you look at Walmart or you look at some of the other big customers or big reference points where they're might using it for Swift, for block object storage or they might be using it for neutron networking for software defined networking in a production environment. So when I say that, you know, some people are saying, well, it's not production ready. I don't feel that. I definitely feel it's production ready. I think there's certain situations or use cases where it may not be quite there yet but in cases like Walmart Labs, I mean, obviously it's clearly in production. And, you know, Morantis obviously has a lot of references to backup that, you know, they've done production ready deployment. So, you know, I thought I'd say about that. I think Jeff, what he was mentioning was it's more project to project basis, for example, e-commerce. But again, you're absolutely right. It's also about people and process. If your people and process are not there, you can go, it'll still be in the innovation and never move to production. So it's, again, people process. The number one thing that drives me crazy, because I think you already know the answer to your question, right? You got to work with a vendor and or build the expertise inside, right? It's like asking, is the Linux kernel production ready? No. You've got to go compile it and configure it and do stuff with it before you've got OS, right? But the other thing is that I find that customers have no idea how the hell scale works. They're like, oh, I've got OpenSack running on five servers on that Netgear hub over there, five laptops. Let's scale it up. Okay, well, what does the switch fabric look like when you get to five, six, 10 racks? They don't know. They scale up to one rack and then they get to the next one and it starts turning into a shit show pretty quickly. And so that's the thing I was saying earlier about the complexity in private cloud like storage and networking and service. Now that stuff's free. You can't go sprinkle software and magically it all manages itself. I mean, this shit is hard. You got to work on it, right? And either your vendor's doing a bunch of heavy lifting for you or you're going to figure it out yourself. I lost a deal for Symantec. I don't know if they're in here. And I was annoyed because I was like, look, you guys are going to have to do this, this, this, and this. And they're like, ah, no, it'll be easy. Nine months later, I went back and they're like, Randy, you are right. We had to do this. I'm like, you rebuilt my product internally. Why did you do that? Why didn't you just take my product? So a lot of it's in about expectations and workloads. Absolutely. Is what are we going to do with it? It's not a generic tool to go everywhere. That just comes back to the case of, you know, it's that nice shiny bumper and they're just like, hey, I've been hearing all this stuff about OpenStack and I can scale. They don't know the difference between horizontal versus vertical and they great here. We'll deploy it in five days and they say, okay, I don't know how to use this now because it's API driven and I'm more of a bottom-up type of scale organization. I still don't get those five days, man. I work in IT organization. You can't do shit in, I'm sorry. You can't do things in five days. Thank you. Who says that, man? Yeah, there you go. Yeah, sure. Yeah, it will be sitting in innovation lab for a long time if you don't have the people and the process and integration for the second day operations, right, without the second job, without your CMD integration. What's going to happen? Those three over there, they just set up. 22 racks for Walmart in three weeks time, including user acceptance, testing, and all of the burn-in and multiple reloads of the software, three weeks time, 22 racks. Yeah, good job. So, you know, I've been in the technology industry for a long time, you know, a lot of gray hair and all that and, you know, unfortunately, a lot of customers look at, you know, many of these technology is like a silver bullet and really it's not and I think it's wrong to say that it's production-ready. It's wrong to say it's not production-ready either. You know, I think it really depends on kind of the workload you're looking for. I think I'm kind of agreeing with you, I'm kind of agreeing with them, and I'm disagreeing with them as well. So, it kind of depends. So, on that note, time's up. So, first of all, I want to thank the panel for sharing their work with us. And we also want to thank all of you for listening and some of your fast questions. So, before you go, we're going to go ahead and raffle off to John Griffith for throwing rocks. So, oh, we're raffling now. Oh, shit, I get to do it. Wow, sweet. Okay, who wants to pad my pocket? I can pick this up. I think he works for EMC. Fantastic. Hey, that's a set up. Okay, so thanks, everyone. Again, I encourage you to go to Randy's talk, stay at the stack Wednesday at 4.50, and also the 5.20. 5.20. 5.20. And then the EMC booth will be open tonight. Sorry, tonight, P-13, I think. That was Brian Charles. Yes, one moment.