 I've been a 19-year HP employee, been writing software a really long time, been working for HP obviously a really long time, you already heard my fire story, but I started HP in 1993, was fortunate enough to sort of catch on the early web craze, worked in an enterprise support organization that turned an old dial-up service into a web-based service so that people could self-solve like the Fords and the Boeing's of the world that would have support contracts with us instead of having to call someone, they could go to the website, get access to information, solve some of their problems. My first web application went live in production in January 1996. My manager would not let me use HTML tables to position any of the elements because it was not yet part of the HTML spec. So it had lots of pre-tags all over the place and therefore would only work with a particular version of Mosaic. So I wrote that kind of wave for a while inside of HP, ultimately becoming the HP.com chief architect. So I did chief architecture at HP.com for about six years, was involved in our, you might remember about five or six years ago, we had a big IT transformation set a project where we took 85 data centers in 26 countries worldwide, consolidated that to six data centers all in the US. So I've been around IT for a really long time. HP Cloud Services, as you may know, big sponsor of OpenStack. We've been doing this a couple of years now. I was employee 37 of HP Cloud Services and earlier this summer I joined the enterprise sales team. So having been an enterprise chief architect, I kind of know some of the pain that folks I'm going in to speak to now, what they feel in trying to design solutions for what it is they're trying to do. So as part of that, these are, you know, big Fortune 500 companies that I talk to you on a daily basis. I hear a lot about what they like about OpenStack and what they think is missing from OpenStack. So when I had the idea for this talk, this is really think of this as a conversation starter. The idea here is that think about what enterprise adoption of Linux would be or what the Linux landscape would look like with that enterprise adoption. If it's just my grandmother running Ubuntu desktop on her laptop, not nearly as many people would care as if, you know, Ford or Boeing is running their mission critical apps on something like Ubuntu or Red Hat or whatever that might be. We care a lot about these things in the Linux world because big companies are using them. So why should OpenStack be any different? Ultimately, to get the foothold that I think we all want to see OpenStack have, we need to get enterprise wide adoption for it. So in the daily conversations with, I have with those folks, I hear a lot of here's what OpenStack is missing in order for me to take it seriously or for me to begin to adopt it in any kind of rapid fashion. Like I said, don't think of this as like a sermon or any kind of edict, but as a conversation starter that ultimately could influence the blueprinting processes that are going on during this convention and in other rooms throughout the week. So first item, which you can't see much of, but just that I remember what all eight of my items are, user management. So Keystone now has the ability to have multi-tenancy. I mean, I have one account. I have multiple users in that account. And enterprises think that's awesome. They don't like the single account per user model. But what they'd like to see to extend that is the idea that I'm already managing users and groups in either active directory or maybe in some other kind of system that offers LDAP as an extension point. What I want to be able to do as an enterprise, I already have my organization mapped out in that way. I just want to be able to synchronize that with the OpenStack, whether that's a public cloud like what we're running or for your own private cloud. So they just want some kind of synchronization. It doesn't have to be even a synchronization that happens all that often. It's not like they're adding users and groups every 30 seconds. This is like a maybe every day or every 12 hours kind of thing that they need. And I know this kind of thing has been talked about on the Keystone roadmap. That's the first example of the kind of thing that enterprises would like to see. Along the same lines, single sign-on, if you have all these users, like to be able to federate those through OAuth or some other standard mechanism such that if I've got a team of six SAs that I would like to grant permission to use an OpenStack-based offering, I don't want them to have to remember all kinds of different usernames and passwords. I want them to be able to use the same authentication mechanism that they use for their normal corporate day-to-day life. Storage monitoring. SWIFT is by far the biggest adopter I've seen so far in enterprise engagements. It's the most battle-tested of all the OpenStack offerings. And folks are starting to use that pretty often to dip their toe in the water using it for two or three storage in a much larger storage offering. But the thing that some folks are complaining about are the monitoring of that storage. If you want to know average age of objects in a container, you want to know things like average size of objects in a container. You have to go and query all those things individually, put them all together, and run your own analytics on that. The kinds of tools that these guys are used to in their Tier 1 and Tier 2 storage, this is built-in. They already get this. They'd like to use the same kinds of tools they monitor for their Tier 1 and Tier 2 storage for SWIFT. XML. I was going to take them at the end because I even have a slide that says what do you think, and then that prompts this whole conversation thing going on. But thanks for the interruption, just the same. So XML support. So JSON is often the first thing that we think of when we talk about interacting with the web services that are all part of OpenStack. And you can just barely see it there. I expected some snickering from the audiences when I described JSON as cutting edge. I really wish I hadn't described it like that. Okay, so I'm just the idiot up here talking. So yeah, cutting edge JSON that everyone's talking about. XML is what, both from a skill perspective and a tool perspective, that our friends at Enterprise have invested time and resource in. A lot of these guys, I can tell you, are, like, they think still that, like, whistel and soap is just the greatest thing ever still. So in order to carry that kind of stuff over into an OpenStack world, a lot of time you're asking those developers to try to learn some new way of doing things. And they're typically not super excited to do that. The kind of folks that are in this room that are eager to learn new things and adopt new technologies are not typically the kind of people that are sitting in IT shops and enterprises. You say that to the HP guy. You're totally not getting that NVM. This is so far totally the best thing you've seen with all this technical difficulty, though, right? So next one. So the Enterprise developer is typically, like, a Java and .NET shop. I can tell you from having been there in the mid-90s to convince the CTO and CIO that this bytecode thing was really going to take off and that JVMs weren't, like, the evil end of the universe. Once they've adopted this mindset and technology, it's difficult to get them to think about it otherwise. So there is some support in, like, the JClouds community. It's done a really nice job of augmenting what we have as part of the standard set of interfaces of language bindings through OpenStack. But we need more of that because these guys are deeply entrenched in these two mindsets. And without this, they're going to think that they have to go retrain their developers to go adopt OpenStack or interact with it in different ways. OpenStack for dummies. I saw Ann Gentel walking around earlier. I don't see her in the room now. But if you guys have worked with her, she's done an awesome job at putting together the team that provides the documentation for developers to use OpenStack. If you've played around with the Swift REST API and you've seen the differences in the troubleshooting and example section of that documentation, you're welcome. I did that a couple months ago in my loan OpenStack submission process. So for developers, Ann's team's done a great job at providing the kinds of documentation we need to educate people on how to use this stuff. But for this guy, and that's supposed to be a pointy-headed boss from Dilbert, it would be really much more awesome if you own it. It's not a documentation that's not just focused towards the developer, but to that IT project manager and trying to educate that person as to what is this OpenStack thing and why should you take advantage of it in your project teams. Along similar lines, we've got to go up the chain as well for the vice presidents of the world. So why should they take advantage of this OpenStack model along the lines of, you know, just as we were talking about during the initial technical difficulty of the session, that there's 180 companies out here that are actively involved in the innovation engine that is OpenStack. There's all kinds of costs to be cut for that vice president and those sorts of things. So covered some technology things, covered some documentation things, some technology things that have been talked about, that have not acted upon in the blueprinting process, some things that are a little bit new. Now like I said, I don't want this to sound like an edict or a sermon, but what are you guys seeing? In your things or your interactions with people at conferences, what are people saying about OpenStack for big companies that they feel might be lacking or could be tweaked a bit to better serve the overall community? Sir, workloads are, oftentimes they have to hire people to get that virtual infrastructure even if it's on a private cloud to get stood up in a way that they can use it based on, integrate well with their existing either monitoring or orchestration or, you know, a whole host of tools that they use to manage their physical infrastructure or even their virtual infrastructure today. So that's like, you know, the LDAP thing is a big deal for them. That's not, you know, that's not something that they necessarily need developers for, but they don't need sysadmins taking manual time to go create manual accounts over on even your OpenStack private cloud just so, you know, that essay and other reasons I wanted to bring this topic and this talk up is and I get this from my own guys that are far more active in OpenStack contribution than I am. The people developing OpenStack if those are the only people that are ever going to use OpenStack then we're screwed, right? Because I mean, there's this number of people that know how to contribute and build OpenStack, but there's this number of people who could potentially use it. And there are tools and documentation and some of these things I'm trying to point out that are missing in order to lower the bar so that those people can start to use it and not just use it but use it in a way that allows them to continue the tools and processes that they have in place today. Sir, I think it is, and I think this, we have two more over here, let's pick a number between one and five because I didn't see who, I brought up Swift monitoring, but you're talking about low level NOVA monitoring and notifications so that, I mean that kind of goes into that gentleman's business continuity comment as well. Can I try to repeat that just so we make sure, is why would you want to build a NOVA cluster greater than 500 nodes? That very much strikes me as the Gates comment of why would you ever want more than 64k of memory, right? It's the exact same thing. It's scale, it's monitoring, it's leveraging into existing system administrator tools. Now let me logistics check here real quick. Now this gentleman was here in the front, could you guys in the back hear him at all? Or would you just, okay, so the acoustics in here are good enough that way that I don't need to police people to the prefer to sit where they are. Now you sir, I punted you for a minute there. And some of that, like the comment was, so that you have to know how to compile all the, you know to deal with all the Python in order to even get the friggin' thing to run, right? See Tom, he made me look bad in a way where you tried and didn't work where, that's awesome man. And like I said, the point of the talk was to spur exactly kind of conversation we're having in the room now. But the challenge going forward is to not let the conversation die in the room and to take this thinking beyond the developer who's gonna self-install. So that we can get into the blueprinting processes and the development process and some of these ideas that we're talking about here. Let me time check here a few minutes after. So even if, even if we had stayed on schedule, we still have about another ten minutes. If we want to take it, if we don't, then we can give away the laptop. But that might be a good segue to stop the conversation and to challenge each of you to bring this back to the development teams. But if we want to talk some more, I want to facilitate that as well. Yes sir. From things I see, right, in, especially in enterprise environments, it's, you know, an upgrade process that you can back out of for one of several reasons. And one that you can do while you're still running, that you, you know, while still flying. I didn't happen to see that demo this morning but are there others that have comments? All every six months you want to keep it in place. In concept I completely agree with you. Except to say that in some enterprises, they just, they fear any change. It is, it's in some places unfortunately a cultural reality that you want to say. It is relatively early. But and that's not to say that it certainly can't be useful. We certainly think so. Or else we wouldn't have it deployed in multiple data. The guy that stands between you and lunch. So thank you for taking the time. Like I said, the challenge is to not let the conversation die in the room but carry it forward into the blueprinting process. I have to show one of these slides that didn't get showed so that I don't get in trouble at home. That's the cheesy version of this slide but my ten-year-old daughter's been watching original series on Netflix. And when she heard the title of my talk, she said, Dad, you have to get a picture of the starship in there. So you guys might think it's cheesy but she's going to choose my rest home someday. So I'm going to I'm going to get in good with her if that's okay with you guys. Okay so can we raise our hands again if we're HP employees? Oh now this time if you're smart you're not raising your hand, right? Okay so underneath your, underneath chairs, we're giving away these circular stickers. There's a partially stuck sticker on the bottom of one chair and if you have that one with some extra writing I have on here so that you cannot cheat and I happen to remember that this gentleman here is where I put it. For those, we have more stickers. We have more stickers and sir you have my sympathy but I'm afraid I don't have a laptop for you. So sir if you want to come on up, thank you so much for coming up. Don't let the conversation die.