 All right. So I think we're going to go ahead and get started so you guys having a good session so far? You exhausted yet? I think I am, like, Thursday already, isn't it? Okay. All right. So, all right. Let's just kind of, we'll start with the introductions, the obligatory introductions. It's me, I think, been a while since I looked like that. My name is Nick Chase, and I'm from Morantis. You may have seen us around, a few people wearing Morantis t-shirts, maybe you've seen them. We are the number one pure play open stack vendor in the market, which basically means all we do is open stack. We don't do operating systems. We don't do hardware. We don't have a dog in any other races. It's just open stack. So we tend to think about it a lot. I am the technical marketing manager over there, which basically means I'm the guy in the marketing department who knows how all this stuff works, which is always good. It's always good if somebody in the marketing department actually understands it. So that's my job. I'm also the editor of OpenStack Now, which you may have seen. It's a weekly newsletter of all the news that goes on in OpenStack. So I do a lot of thinking about all this stuff, which kind of led me to the subject of this talk. So let's just kind of talk about where we're going here. We're going to talk about hybrid cloud today, but we're going to talk about a very particular idea for hybrid cloud. So we're going to start off with sort of defining what IC hybrid cloud actually is. Because everybody's got their own definition. It's kind of like cloud itself. Then we'll talk about the sort of levels of hybridization, you know, because what some people are calling hybrid cloud, maybe not so much, versus where I think we're ultimately going. Then we're going to talk about this idea of OpenStack as the new web. And I'll get there if you're looking at it going, what the heck is he talking about? Don't worry, I'll get there. And then we'll talk about the sort of steps to creating what I think of as the worldwide cloud. And of course there are some technical hurdles. We'll discuss those and what we as the OpenStack community would need to do in order to make that happen. So let's kind of start out by talking about what hybrid cloud is. Everybody has their own definition. But really at its heart, hybrid cloud is basically where you have multiple clouds. Usually there's private cloud and there's public cloud. Okay, yeah, all right, but it doesn't have to be. Just means there's more than one cloud. Might be on different systems, might be on different platforms. Usually it's in different locations. And they talk to each other in some way. Or they interact with each other in some way. Sometimes the job goes over here, sometimes the job goes over there. That's basically the idea. It's hybrid in that you've got lots of different systems all working together in some way. Thing is, if you actually get down to how that works, if you ask anybody, it's usually something like this. Maybe you've seen this cartoon occasionally. 20 years ago when this cartoon first came out, I got a t-shirt that said that. The then a miracle occurs. That's the step where we're at right now. Everybody, you know, there are lots of vendors that say, we have hybrid cloud. You can have your on-premise and your off-premise. And Morantis is one of them. We just announced, you know, Morantis OpenStack Express, data center on demand as a service. And that's great. But I want to set your expectations here in this talk. We're not going to solve the world's problems in this talk. We're not even going to solve OpenStack's problems in this talk. In fact, this talk is going to be a little bit like a design session in that I'm going to present to you some ideas on where I think this technology is going. And then I'm going to say to you, hey guys, what do you think we should do here to make this happen? So I just kind of wanted to lay the groundwork here and let you know where we're going. All right, does anybody have any questions so far? All right, keep going. By the way, there are two microphones here. So if you do have a question, please feel free to run up to, well, don't run because, you know, it's Thursday, we're all tired. But feel free to hit up to the microphones. OK, so hybrid clouds themselves, typically we hear about a couple of use cases. We say, oh, OK, well, you know, we do, you know, dev tests. We put our development work on one and, you know, our tests up on the public cloud and then we bring it in-house when it's ready. Or we use it for cloud bursting. You know, we have our normal loads in-house and then, you know, when we need extra load. We, you know, outsource the rest of it and so on. Or we use it for better resource allocation. And these are very well understood ideas, which is great. That's all, that's all nice. This is so new. What haven't we thought of yet? And I'm going to touch on that more in a little while. So let's talk about the sort of stages of this sort of hybridization that I'm talking about. And this is, I think that as we go along, what we're going to find out is that different systems and different organizations have different levels of hybridization. And they follow along this sort of continuum. We start out with isolated, I'm going to go over what all these things mean in a second. So you have isolated clouds, pardon the weather reference, it's just unavoidable, I'm so sorry. Homogenous clouds, heterogeneous, and then this worldwide cloud idea. So let's talk about what I mean by each of these ideas. So isolated clouds. So we've got, you know, we've got, maybe we have multiple clouds. Maybe we've got an on-premise private cloud. Maybe that's based on OpenStack and then we've got a cloud over here at Rackspace that's based in OpenStack or we've got a cloud over here in Amazon or we've got a cloud over here in Google. But we've got multiple clouds but they're not really talking to each other. Or we have a hybrid cloud. This is a situation where a lot of enterprises, they say, we have a hybrid cloud strategy. Yes, they do have a hybrid cloud strategy. They don't really have a hybrid cloud because, well, we're going to manually decide, well, we're going to put this job over here. And we're going to put this job over here. And there's nothing wrong with that. It is, in fact, a hybrid cloud strategy. That is correct. But it is just the isolated form. Then you've got homogenous clouds. This is where everything is, all of your clouds are of the same type. So maybe you have an OpenStack private cloud. You've got an OpenStack public cloud. Everything is using the same API, which is great. So the software that you're using is going to assign your jobs automatically. Terrific. That's a sort of homogenous idea, a lot simpler. Then we move up the stack to heterogeneous clouds. This is where you've got multiple components, multiple system types. Maybe I've got an OpenStack private cloud. I've got some stuff on an OpenStack public cloud. I've got some stuff on Amazon. I've got some stuff on something we haven't even invented yet, something like that. But everything's managed by one client. The person who is starting up a job, they don't know where it's going. They don't care where it's going. And that's good. That's the heterogeneous level. And then there's this worldwide cloud notion, or the sort of cloud of everything. I'm a little bit mad because Cisco stole my intercloud reference, but what can you do? So this is basically the idea of the heterogeneous cloud. But there's a single management sort of grid. I'm going to assume it's OpenStack because, well, where are we? Also, it's the best. But it's joinable by anyone. And that's the notion that I want to sort of bring into this, where anybody can join the worldwide cloud. I have an extra hard drive. I'll make it available. I have some extra compute power. I'll make it available for somebody to throw a VM on. I have some extra whatever. And I can make that available. This is the ultimate goal. And that is what I am calling the sort of ultimate hybrid cloud. This is the worldwide cloud level that I think we are all heading for. That's what I think we need to get ready for. So let's talk about why I feel that way. So we talked a minute ago about these sort of existing current use cases. Dev tests and cloud bursting, we talked about all those. We know those. We hear about those all the time. If you've attended any of the other hybrid cloud presentations, you've probably heard all of this all day. And there's nothing wrong with that. That's great. But what haven't we heard of yet? I mean, we can't sit here and think OpenStack is, what, three and a half, four years old? We already know everything it could do. Are you kidding? No. Of course, we don't know everything it can do. We're never going to know everything it could do. I mean, I think we may be asking the wrong question. We keep saying, is OpenStack the new Linux? What if we're wrong? What if we should be asking, is OpenStack the new web? Now, of course, you don't develop these presentations in a vacuum. I have run this by several people. And every one of them has looked at me like, what? They've all looked at me like I have three heads and the middle one is on fire. But let's think about this for just a minute. What did the web give us? Well, it gave us democratization of content production and use. Anybody could go ahead and do this. So we created new ways to communicate. Twitter and Facebook and communicating by companies just going out and putting out their website. None of that existed. If you wanted to know about a company, you had to call them up on the phone. You couldn't pay your bills online. You couldn't find out what your friends were doing. You couldn't do any of that. But it started out. How many of you are old enough to remember the web starting out? So yeah, I like this room. There's people here who remember. You remember when it was all about cats and pictures of cats and what is my cat doing? Listen, I love cats, but come on. Remember we used to say the great thing about the web is anybody can put up a web page? Problem with the web is anybody can put up a web page? That's where we were. Ooh, where'd I go? But new use cases kept being dreamed up. I myself worked in the web on the web. And I started on the web in 1995 when we were looking and going, hey, you know, we could update a database from the browser. It was like, we can? Yeah, we can. It was an awesome idea. And the reason that we were all able to do that is because everything was open. HTTP was open. Anybody could make a web server. Anybody could get a web server. If you had a web server, you could go to the W3C.org page. Find out how to write HTML. Anybody could do it. OK, great. And so anybody did do it. And the barrier to entry was so low that we had all kinds of ideas coming up. All right, what can opens that give us? Well, democratization of resource production and use, new ways to process data, new use cases, never dreamed of by its founders or us or anybody using it now. All based on open standards. Anybody can do it. Anybody can go get OpenStack. OK, you've got to work a little harder to install it. But there's lots of good distributions out there. By the way, Miranda's has one, software.maranis.com. But you'll notice, I mean, look, there's not a lot of difference here. It's not such a crazy idea. Yes, sir? He asked me if I could go back one slide to here. I have it in quotes. And let me tell you what I have in quotes. Well, let me just repeat that for the camera. What he was suggesting is that I take off these quotes on this page, but leave them on this page. The reason that I put them on here is because actually, they're not standards. HTML is not a standard. It is a de facto standard. But technically speaking, it is not a standard. It is a recommendation. HTTP is a standard. OK, I'll give you that. You're right. I'll get you on that. But I was thinking of HTML. So we can split hairs all day, but that's why there's quotes there. All right. So all right. Some are standard, some are not. Can we agree on that? OK, thank you very much. So that is why I'm thinking, this is not a crazy idea. How do we get there? And I think we get there in stages. I think we start with where we sort of are now, for the most part. And this is where we're going to get a little bit vague on the technical side. For those of you who live and die by Garrett and Jenkins, bear with me for a few minutes. So basically what I'm thinking here is, step one, you have a one cloud. You have it's private or it's public. Doesn't matter. It's controlled by an API. Notice I have here, not AWS. Now look, I'm not digging on the people who think we should support the Amazon standards, the Amazon API rather. I myself was one of those people until the court came and said that you can copyright an API and make people pay you to use it. But now that they've said that, sorry guys, I've got to jump in the note, forget it, stack. So it's step one. You're choosing manually where you're going to put your resources. This is the hybrid strategy level, the isolated level. Here you have all of your resources are on the same network. Everything has to hit Keystone. Everything has to access MySQL or whatever it to be or whatever database you're using. Step two, you have different clouds. OK, when I say different clouds, I mean different conceptual clouds. Maybe they're separated by availability zone. Maybe they're separated by some other open stack mechanism. Maybe MySQL is replaced by some sort of distributed hash table-ish method since now we're talking about having resources in multiple locations. Maybe the clouds are connected. Maybe they're not. Maybe the worldwide cloud consists of this giant cloud over here and another giant cloud over here and maybe a little cloud over here that doesn't talk to anybody else because it's anti-social whatever. At this level, I'm thinking multiple contributors. This is where you say, I've got some extra resources. Let me make them available. But each cloud may have its own resource policies. Maybe I don't want to take just anybody. Maybe I'm going to create my own cloud. I only want to accept people that I know. Maybe I'm a big company. I only want to take resources from other big companies. Maybe I only want to take people who are vetted. So I know that I'm not taking in a money laundering organization or something like that. Step three, I feel like I'm in the producers. Step one, find the worst play ever. OK, step three, this is where you get the point where you've got all your clouds built. Now you start to have things like services built on it. Maybe you've got Sahara for big data processing. Or the Morano application catalog. I am so excited about the Morano application catalog. I'm going to just kind of digress for just a second. I love this thing. I was involved in writing this back. I helped them with this back. It's so cool. Different clouds can have their own catalogs. They can decide different policies in billing. And you can deploy your applications just by dragging things around. And there you go. I totally recommend at least looking into it. But at this level, you've got these resources being spread around in a way that is available. So technical challenges. So if we've got people coming in from all over the world, does everything have to be on the same network? Can we use something like OpenVPN to create this network? What do we need to do this? Does SDN come into play? What about NFV? There are all kinds of challenges in this space that we're going to have to overcome. Trust. This is really the big thing in a lot of ways. Users want to know where their resources are going. Is it going to a Rackspace server? Is it going to an Amazon server? I don't care what the protocol is underneath. Is it going to Joe's basement server? Joe may be a great guy, but maybe I don't really trust his household internet connection to make my stuff available. On the other hand, if I am making my stuff available, maybe I want to know what's going on on my hard drive. So do we want to deal with that? Can we solve this problem by nova flavors? Maybe there's a nova flavor that shows this is a Rackspace server, this is an HP server, this is a whatever server, and so on. Do we use something like graffiti for additional metadata that allows us to get all that kind of information? Billing. Let's be realistic. People are going to do this because they want to make money off of resources they already have. I mean, that's how virtualization started. Companies said, we got all this resources. Why are we wasting it? So I got all these resources. Why am I wasting it? But I'm not going to just leave it out there for no reason. I want to get paid. So we need to be able to handle that. Coordination among services. Does the underlying machinery matter? I don't think it does. I don't think it matters what kind of cloud it is. I think everything can be linked together. Somebody, I think it was RightScale, had a sticker, a couple of things in Portland that said, all your cloud are belong to us. There are services and APIs and things like that out there that are making moves towards connecting all of these different clouds. Jumpgate, an API from IBM and SoftLayer for connecting different clouds. Their whole intercloud thing. I don't know if they have any actual working code yet, but they're doing a lot of talking about connecting different clouds of different types together. But at the end of the day, users, when you go to a website, do you know whether the HTTP server is Apache or Nginx? All right. Maybe I'm asking the wrong room that question. In general, most users don't know and don't care. All right. Traceability, all right? Let's be realistic. NSA aside, some users still want to maintain their anonymity. On the other hand, there needs to be a way for things like Digital Millennium Copyright Act requests. Maybe I don't want to make it known that this is my hard drive out here, but I still need to have a way for somebody to say, hey, you know you're streaming frozen from there, right? So stuff like that. It's harder to go after individuals, but on the other hand, it's harder for an individual to fight a subpoena. So there's a lot of trade-offs going on here. In terms of what the community needs to do, there are basic security issues that we need to resolve. You've got strangers joining your cloud. How do we do that? Do we create multiple levels of security? Do we create one-way credentials? I mean, we don't want to give everybody access to the MySQL database that's running our cloud. That would be nuts. Do we use federated identity? How do we do that? We need to improve cloud-to-cloud communication. We need to make it easier to move workloads between clusters. We need to make it easier to share resources from multiple clusters. And we need to standardize the OpenStack API. That's where we started this conversation out, remember. It was a standard. HTML was standard. Sorry, I still have to give it the air quotes. It was a standard. Everybody knew what to do. Everybody needs to know what to do with OpenStack. Whether OpenStack is running on Rackspace, or HP, or Oracle, or Mirantis, or whatever, it's all got to run the same, OK? We need to develop relationships with projects like Jumpgate, or anybody else who is consuming these OpenStack APIs. We can't ignore them. In terms of where we go from here, how do we get to this picture where everybody can contribute? How do we get to this hybrid nirvana? And that, surprisingly enough, brings me to the end with 13 minutes remaining for questions. So does anybody have any questions or comments? Yes, sir. There's a mic back there. Hello. Thank you very much for a great presentation. I have a question about storage. It's a question and a comment at the same time. If you are lucky enough, you probably will not care about, or your application will not care about location of your storage. But for majority production environments, you really care where your data is located. So if you will start off loading anything to another cloud, you need to ensure that you have the same data or have some storage policies. What do you think about this one? I think you're right. I think that that kind of policy is important. And I think that that's the kind of thing that can be handled in terms of specific cloud policies and types and things like that where it's known what it means. When I say I want this kind of drive, right now we're saying, oh, give me an SSD drive. Maybe I don't want to say give me an SSD drive. Maybe I want to say, give me an SSD drive located on this cloud and this data center from this company. What do you guys think? If you're talking about new resources on this, for sure. But if you already have your farm of applications, whatever running with your existing data, you just cannot do right now allocation anywhere else because you need locality. Sure. But if you will do any sort of replication of your data between multiple clouds, then you probably will be able to do it. Right. And that's where it comes into play. Yeah. Thank you. My name is Amir Tapp, dealing with Cypress Consulting. Great presentation. Thank you. Going back to the question of standards, I've been listening to what you were saying, and I think your reference to HTTP HTML is a wrong analogy. Because if you look at the way the networks grew, and my background is in service provider networks, I look at it from that point of view. What made everything work and what made web work is actually the way to connect networks. And for a long time, you could go and have separate networks that don't talk to each other. And you can actually have HTTP or HTML, and it wouldn't be no good. What really made it work is BGP, right? Yeah. And what did BGP do, right? I mean, if you have network people that tell you that we don't want to mix blood, right? That's the first thing. Even within the same corporation, I mean, we work with some big service providers. Even within the same company, they don't want to mix blood. So what does BGP give you? What BGP gives you is a way to enforce your own policies, be able to exchange information in a summary format, and be able to exchange incremental change in a scalable fashion. So first, I'd like to say, I think it's a great presentation, but you should think about how networks grew to enable internet. And that was not because HTTP was a standardized or HTML was there. I think that's all good, but it was BGP that made it all work. What we really need is a standard that follows that model. And once you have that, then you can get to something like this. I think you're right. I think that, and I appreciate that perspective on it. I think that, yeah, I mean, I was looking at it from the user perspective, but I think you're right. We have to look at it from the behind-the-scenes perspective, especially since we are working behind-the-scenes. So we need to look at, well, what do we need to enable those kind of policies? And there's a lot of policy work going on right now. There's Congress, there's, I can't think of the name of it that's competing with Congress, but, and so on and so forth, but you're absolutely right. We do need that kind of policy work on the back end to make sure that we've kind of covered the bases in that respect. So we have other questions or comments or we see where we are time-wise. Well, I guess if there's no other questions, I think we can give you 10 minutes of your day back. All right, well, that's it. Thank you very much, everybody.