 So, I was very excited that we were able to pull this panel together. We've been talking all week about all the big users out there and some of the different industries they're in or sort of scientific computing or big businesses. And now we're really looking at the diversity of OpenStack deployments geographically and hearing from some of the leaders who are out there building and deploying OpenStack clouds all over the world. So, that's what we want to talk about today. I'm Mark Collier and with the OpenStack Foundation and we're going to give everyone a quick chance to introduce themselves and tell us where they're from and what their company is working on. And then we'll go through a few questions. My name is Mauricio Rojas. I'm working in Mexico. I'm Chilean by the way, but here the company that I'm working today, they take me from Chile to start working has a CTO. Well, what is doing Kio? Kio is a 100% Mexican company. We have 14 data centers around Latin America. In Mexico, Central America, we're starting with a couple of data centers in South America. We are a company of 1,500 people, it's very big. And today we have a lot of cloud services for Enterprise. Enterprise is our main target today. What we have there, we are working with government, financial, mostly of the company are in this industry. That's all right now. Thank you. Hello everybody. My name is Mariano Cunietti. I'm in the CTO at ENTER. ENTER is an ISP based in Milan, Italy. So as an ISP we do mainly two things, which are hosting and connectivity. We sell fiber, copper, XDSL lines, MPLS networks across Europe and the world. And we also provide enterprise services based on virtualization or physical hosting or housing, traditional services. We developed a cloud on OpenStack to provide VPS services. The name of this service is CloudApp, CloudApp.it. And also another service named Subserver.it, which is the same as CloudApp, but with pre-configured packages, software pre-installed on top of it. So my name is Brian Aker. I'm at Hewlett Packard. My title is like fellow and there's a bunch of other stuff at the end of it. But essentially I own managed and platform services components that run on top of HP Cloud, which is entirely, well, not entirely, mostly, all of OpenStack. So we actually have Nova and Glantz and all that being publicly available. I'm Nachi Ueno from Entity Japan. I'm mainly working for the development side, so sorry, I'm not marketing guys. Okay, I can translate it into marketing speak. Yeah. So we are working for the OpenStack from very early stage and sending tons of patches to the communities. I'm mainly working for the quantum and I'm one of the core developer of the quantum now. So basically we are preparing some OpenStack cluster now. Thank you. Thank you. Great. Well, I thought it would be interesting to hear a little bit about the conditions in your local markets that you serve in terms of what the state of interest is in cloud and how much adoption there is and what is it about your market that might be a little bit unique in terms of OpenStack interest or cloud in general adoption? Well, in Latin America, the cloud services market is not digital yet in share of the global services, but today we can say we have roughly $300 million as a cloud services market, but we have I think the highest growth rate if you see the breadth of the continent of the countries, whereas there is a very high... We are not prepared if we think about the companies of the user in order to work with automation, orchestration, all these kind of tools today. The customers think that a bunch of virtual machines is a cloud and it's difficult to explain them that we need to go to have better control of the virtual machines or to try to provision faster. The main challenge is to teach them about this. Kio is working really hard to educate the future users, our customers, about this with different seminars on training in different countries of Latin America, but this is the hard part. The other thing is the public cloud is more related to SMB. It's not so enterprise today or mid-market. We are working hard to do that, but the mid-market and the enterprise is just putting focus on private cloud. They want their own dedicated infrastructure to do that. They want to sometimes have the control, but they like to blame somebody about the problem that they can have in their services and today this is the reality that we have in our market. Thank you. Related to Europe, the total amount of cloud expenditure is 40% compared to the North American one, but the investments in private cloud are exactly the same if not higher for Europe. When it comes to Italy, we are talking about a 19 billion euros market, which is decreasing by 3% last year. Instead, the cloud market is growing by 57% and for a total amount of 620 million euros amount of the total market. When you talk about cloud in Italy, mostly you talk about YAS and SAS. YAS actually is not cloud as everybody knows it, but is intended as VPS mostly. We have one of the largest YAS providers in Italy that's advertising services on the TV, so its process of commoditization is almost ongoing. When it comes to SAS, we are talking about mostly mail, mail services, storage services and disaster recovery. There's another important point about Europe and then about Italy that is the privacy regulation. I've been to the European Cloud Conference last month and I've attended to a speech by Mrs Viviane Redding, she's the commissioner for human rights and citizenship and she's also the vice president of the EU and it's quite clear that next year by February the data protection, which is the law in Europe for privacy, which is aged from 1995 is going to be updated by February and some rules will change. Will change mostly for those players that are not EU based. The rule is simple. If you play on this ground, you play with our rules. At this moment, players like Amazon, Google or Microsoft can play with their rules. It's going to change. It's going to change and there's been a quite interesting debate because the European people talked about the difference of right of citizens for privacy to be let alone and the American or the global operators talked about the rights of the consumers. There's been a lot of debate. There's been a lot of lobbying also and in February, I think that the market could open to alternative and local initiatives and offers, so OpenStack can play quite a role in that. Obviously, HP is global so it's billions and all that, so let me put a different spin on this. Here's the average CIO. They walk in. This is, by the way, one of the great reasons why I actually think OpenStack is a big future. CIO walks in and goes, so here's the problem. All this stealth IT has been spending millions of dollars that I've just figured out on X, Y and Z Cloud. How do I bring this in-house? That's great. They want to bring it in-house, so like OpenStack and this comes from across the spectrum right now, so Fortune 100 beyond all that. Everyone is asking the question of how do they bring this stuff in-house? You have local governments being local to say countries who are also interested in this for IT protection and all that. I think, what was it? I talked to, you probably don't care, CTO for World Economic Forum who I had dinner with about a year or so ago and he was like, yeah, there's no way in hell we put this shit up anywhere in somebody else's cloud, no way. Here's the problem that the local governments and everybody else is going to face. Stealth IT means that they're probably going to pick wherever the cheapest rates they can get and they're going to push the stuff out everywhere. Anyway, just to push back on the thought, local is going to work to some degree, but you're actually going to end up with a stealth group who's going to say, well, we could do it locally or we could spend half the money and wait for some regulatory agency to come in and try to hunt us down. Or in some countries, you have people who are activists who don't want it in the local countries who specifically want to put it out there. I think when we look at these stuff, how do we move OpenStack in a way so that when we talk about SOC compliant, HIPAA compliant, you name it compliant, how we actually start baking that into the product because people are going to... There is going to be some localization, some requirements localization. Some companies are going to completely control it. There's a bank in Japan that I've got to visit where it was great. You go into the bank and they've got this sliding panel that moves across out of some James Bond film and they've got a data center that they have this automated robot forklift that brings in computers, drops the rack and when they're finished with the rack, comes in, picks it out and then squashes it in the backyard because they won't actually want the data. But there is no transport in and out of that. So anyway, I think we have to build it into the cloud because it's going to happen anyway. I like the robot idea, so I want to use that. So in Japan, OpenStack is very focused, so we held some OpenStack Japanese as a community event and more than 1,500 people joining. So growing and the community is growing and many companies start considering to choose OpenStack. So this kind of segues to my next question, which is for your companies deciding to build a cloud, why OpenStack? Why did you pick it? Sure, you looked at a lot of other technologies. Maybe you've used them or tried them or... Well, I'll go for it. Brian? I actually came in right after HP had made that decision. So they had actually had something that was built internally, that they were looking at, like, going forward with. And then someone said, eh, bitch, it's the one that would work out better if we actually leveraged something that bigger part of the market would actually do. Which is obvious, at least to me and probably most of this audience. But when you look at it, like, if you look at all the different, like, cloud piece of software, OpenStack actually is focused on being a stack. Like, there's identity as a service, something we don't talk about much, but that's kind of a radical concept. So that's some of the obvious reasons why I actually pick it, beyond, like, licensing reasons and all that other stuff. Like, it truly has something that has a lot more of a vision to it than what we really see with the other pieces or the other projects. So, yeah, that's a simple, it is open. And open communities. So if you are adding new functions, we can discuss them there. But if this is a product, we should ask the, please add this functionality on the next release line and then wait to, or forever, something like that. So then, if it is OpenStack, we can join the discussions, we can propose the functionalities. That's the biggest reason for us. So the ability for you to actually have a direct impact on the roadmap, the features, and basically, if the software doesn't meet your needs, you can get in there and get involved. Yeah, exactly. Versus a traditional software vendor relationship. Yeah. So, prevents some specific vendor locking and the open way and they, for example, the Kanterm or Shinder has plug-in architecture, so many vendors are plug-in. So it has given me many choices of the product. So something that, like, is very different about a project like OpenStack compared to other ones, or specifically not having a company that is one single company behind it, there are some advantages. There's some disadvantages, obviously, but one of the advantages is if you're inside of a company, and obviously I have some history and knowledge of this part, you always end up with this ongoing debate of, like, well, do we create an enterprise version? Do we create a community version? Does this go in the open source version? Does it not? Oh, we just had a leadership change. Somebody else has got this brand fningal-like idea that, you know, we're going to have to, like, kill over, like, how do we keep something proprietary? Like, that debate is extremely boring and it does not actually help. And something about, like, having something that's back in the way that OpenStack is, you don't end up with that debate, and that's good. Or if you end up in the debate, it really doesn't matter because it's often some other company. It's not in the core of the environment that that debate's happening in. Because it's not good. It doesn't actually, like, reflect a good product long-term. And I have experienced that as well. And I think a lot of times it's a distraction, right, because it's a constant, sort of, it's constantly unclear where the line is between, you know, where the innovation's supposed to go and whether making the open source version better is that success or failure. So you're not sure whether what success looks like you're going to be, you know, having a lot of trouble moving quickly. So, sorry, go ahead. Well, in our case, why we chose OpenStack is more related to a necessity of prices. That was the first. Because in the S&P market, all the pricing is an issue because you need to be very aggressive. And we were trying to find a way to do that with normal vendors like Microsoft or VMware. But it was very difficult because the license cost and the support and the services. For us, it was a risk to try to go to OpenStack at, for instance, because you can't find people that can manage this or can't develop that. Since we have a team that is part of the community that helped us to do that. And the main reason was first price. The second reason was time to market. Before that, we were trying to do this with Microsoft, with Hyper-V. But it was very difficult to develop any feature that the customer demands about orchestration, automaticion in S&P. And it's expensive also. Then the time to market that we can get using OpenStack everybody is helping to try to put it this in has a cutting edge technology for cloud services. And that is helping a lot to us to have a nice feature to offer to the S&P market as the main differentiator. I think this is a couple of things that the reason that we decide to go with OpenStack. And the other is we can use commodity hardware, for example, behind this. Was very important for us to try to lower the cost of our hardware. Usually in our company, the 95% of our customer are almost everybody is enterprise, government, financial, with very high cost. And it was a risk because it was to try to change the brain of our operators and our sales guys in order to go and to bid for a new technology in order to attend the needs from the market, particularly of the S&P. Great. Do you have anything to add, or should we? In our case, it's a simple story. In February 2010, VMware changed its policy for the service providers and forced them to move to the VSPP program, which is the VMware service provider program. So in order to install the vCloud director and any other vSphere installation, we would have charged, not by socket, but we would have charged by virtual memory allocated. That would have destroyed any possible income we could have from selling a cloud platform based on that price model. It was a meeting we had with the sales guys in Italy and we discovered that it was nothing to do. We could not change this. Huge margins had instantly became very, very thin. So we had to switch to an open source solution. We evaluated the eucalyptus open nebula two years ago. So there were very, very at an early stage. And then we came to the OpenStack site and we arrived at the documentation, which was at the time very, very complete to understand how the project was built, was designed, was well-ordered. Order was the thing that I found in this project. So we started working with cactus. We planned to go live with Diablo, but then we switched to SX as it came out. And so we went in production with SX and we are running production with SX now. We have plans to do things with Grizzly before summer. But that's it, that's why. So the reason why we have chosen OpenStack then one is economical and it's this, the other one is that we could develop our own product. This is important. One thing that is very useful to reach out for the customer are the interfaces. If you have interfaces, you let the people use the cloud. They don't want to cope with APIs. They want to use the services. And in order to use services, they need interfaces. We have worked a lot with the partners. Some of them are here. Joe Arnold by Swift-Tac is one of them. We have worked with the guys at Scalar because we think that providing interfaces is a key to sell a cloud product. So we have chosen OpenStack because it was flexible to our needs. And the third reason is philosophical. It's open source. So we like it. We want to focus on technology, not on product. We want to focus on our product by using the technology, by knowing the technology. Thank you. It seems like kind of looking for some common themes here that everyone was, I guess, attracted to the licensing model that's free and the flexibility and the ability to influence the roadmap and the ability in terms of flexibility to customize and turn it into the specific product that you're looking to bring to market. So I've got one more question and then we can open up for questions if anyone has any before we wrap up. So in terms of your specific product offering and your business strategy, can you say a little bit about your offering in terms of what market you're going after, whether it's a specific vertical, if you're going after price is the main consideration for the customers you're going after, just a little bit about that approach and what kind of customers you're going after. You mentioned earlier some of the segments that you were addressing. So maybe you just want to expand on that a little bit. Well, yes, we are very happy with the SMB market because it was a nice test of the technology. We are the first service provided doing that. We don't know any other in Latin America trying to do that. Another thing, we have a connection with our marketplace for SMB that make it easier to the end user to provide machines or all this kind of stuff with other products that we are offering today for SMB. We are thinking seriously to go for enterprise but the thing is you have to first, you have to know the culture that we have in Latin America. They don't like to open a ticket for anything. They like something face to face. Sometimes when you want to close a deal, you have to invite them to have lunch and it's easier to talk with them when they have a meeting in their mouth. You can convince them that you have a nice company and they bring you the time to do that. But it's more associated to human relationships. You need to create the link first in enterprise. And the other thing is they like to define what technology we are going to use because there is a very big influence of the other vendors. For example, with BingWare, Cisco or the other ones, the big problem is they bring you the checklist of the technology they want in their cloud. It's difficult to try to change their minds and sometimes you have thousands of parameters of technology that you need to manage. And they like to have this because they like to say, I have BingWare, I have Cisco, I have them. What we are thinking is probably go to the next step with OpenSnap for the mid-size, for the mid-market, is more open to these new technologies and they have a very high necessity to have fast provisioning of their resources like networking, security, virtual machines, all this stuff. But we need to work hard in order to educate these companies and show them how this works. They don't like just to buy something because you put it in your website. They need to see it. They want to touch it. And sometimes you need to take them to the United States to see the brand or to buy them, for example, to these kind of events in order to try to change their minds about this technology. Another thing is availability. Performance is something important, but they have more concern about data availability. They don't last to lose any block. And sometimes it's difficult when you try to bring very nice technology like Snatchos or so on in order to protect the information. Sometimes it's not enough. They don't want to lose any block of data because when you are attending financial and government, a block of data could be, or I lose some records of the social security and could be unused in the newspaper. And it's a very hard way to try to provide this environment, this kind of services. But we are confident that in the mid-market, we can have room to offer this kind of services. And that could be our next step. The thing is we need to prove that our enterprise cloud services based on OpenStack could guarantee some security in data and services and performance. And that is going to be the hard part, to try to promote the services with the customer. And the other topic could be orchestration. They don't like to go to a portal to provision by theirself. They want somebody to do this for them. And that is a lot of cost in managed services that you need to provide and you need to find the talents, the required talent to do. By the way, we are hiring. Somebody wants to go to Mexico, Elena. Elena is our human resource director. You can talk with her and you are going to be next to the Caribbean and the Mexican food is really nice and you are going to like it a lot. It's going to be a nice experience and adventure. In about two weeks from now, you should watch musical chairs as people move around. So the real problem with that, poke at that. We'll have a little graphic like this, a little anime. The real problem with this, by the way, is actually we have to actually train more engineers outside of this pool of engineers and figure out a way to reach those because otherwise, swapping people back and forth, it actually is going to chew up more energy. So the real question is, how do we actually get more people involved? Anyway, just my two cents on it. Cool, go ahead. Our main target, since the Italian market is mostly based on VPS offers, was the typical audience for VPS services. So IT professionals, freelancers, startups. We moved from a very low-skill kind of persona to an even more skilled and now we are developing services for people who are able to use cloud as it needs full power. So I think that, well, in Italy, 95% of the companies are very small companies. And when I mean very small, I think that an American could not understand what very small means, mostly 10, 15 people. So very small. So it's difficult to sell large infrastructures to this kind of enterprises. It's more simple to send the single servers. So my take on this, there's a bunch of enterprise customers out there that we have to actually build a lot more framework around before they can actually use this stuff. I mean, the concept of like, yeah, just use curl hair. That's, there's a lot more that has to be done out there. And there's actually a big, there's like a much bigger audience out there who is actually needs a lot more simplification than what we have right now. So, I mean, to me though, like, if we look at the market, I touched on this in the Keystone this morning, but like any part in the beginning of an ecosystem, there's lots of usual random parts and they don't fit very well and like that's usually what happens. And slowly we see all the baseline stuff all actually just start fitting together and making sense. Go back a 10 year plus years ago, you would like look at people who were like selling, like build a lamp stack kit. Like hold it, doesn't that stuff just all work together? Like why? And we're seeing that same thing actually happen right now. And I believe within the next year, because OpenStack's moving very, very rapidly that we'll see all that base stuff solve. There's gonna be one path for, you know, a happy path for a lot of these pieces. And what we'll see is innovation start occurring more on the outside of the ecosystem. So we'll start seeing that in, you know, paths, managed, whatever you wanna call it. All of those pieces in a lot less churn in that middle, like in that blower part in the middle, like in even the mid tier. So, anyway. So, Nadia, I'll let you off the hook on this one. If there's any, just to see if there's any questions before we have to end. Yes. I think there's a mic here, but to be loud either way. By the way, so what you're talking about, like cost per unit, especially in the lower end, not when we get into like paths and stuff. That is part of operational excellence. So like this is where OpenStack has to be refined enough that you can bring down the cost to where you can do a competition. Cause obviously, right, right. Like what it's taking us to install OpenStack today. Well, obviously Amazon and Google don't quite have those problems at the moment. So to bring down that cost initially, if you want to talk about in the IAS market and so forth, where really that whole market for the most part is going to become free here at some point anyway and everything above that's going to be cost. Like this is the part where we have to get into the installer. We have to really make this thing work, so. But about your question is the service is not free because it's open source. You have to pay engineers to take care of this and you try to be aggressive, but sometimes it's not enough that no pay licenses because you have operation, you have cost of collocation, you have cost of a communication, for example in Latin America, you can find cost of $50 per megapit per second. Then it's a special, I can say, you have to take care of all these costs in order to provide, to prioritize the service in per brick wall machine, but it's not free. You're not bringing these machines for nothing. You need to have money for this. And it's slow, it's aggressive, but you need a team to take care of this. It's not an easy way to do that. About Amazon and the others, today in Latin America, the main industry is government. They cannot take the data out of the country. It's part of the thing that they need to take care. Because this is a big opportunity for us to attend this market, this industry with solutions like this. Any other questions? Yes. That's, by the way, I think the maturity. What he's asking about is the current complexity and all the different options. That's what a little bit of, how many people here were two or three summits ago? How many people in the room? So, okay, it is so different now. The stuff is Diablo took me personally two days to figure out how to install. I got Folsom done in a few hours. A lot of the stuff that you see churn in that lower place is just all has to do with, as the market, as the product matures, you're gonna see a lot of that stuff fall out and value comes off the top, not in the lower level. So that complexity, I'd like to think that this is ambitious and within a year we're gonna see a lot of that just fall out and be one thing. I mean, it's low hanging fruit. That's also why you see a lot of people doing it, which also means there'll be a standardized way of doing it here within a very short generation. So, that gets solved. So, I think that we're out of time for the panel, but I can see you really wanna ask your question. So, ask it real quick and then we'll wrap. So, I'll put on a non-HP hat, or actually, this is actually an HP's best interest anyway. So, the answer to that right now, which is a button I push with the foundation is, the foundation has to create a standard set of API tests that test API and functionality response so that then all of the vendors are actually certified. So, like you shouldn't care if it's HP or insert anybody else because the API is verified. And this is something that I think the foundation, from what I hear from different board members, I'm not the only one who pokes on this, they are gonna do this. So, that's where the foundation has to create an API suite that everybody passes or doesn't pass and you know going into it. And that's the only way I think to make sure that exactly your concern doesn't happen. And by the way, it's not an HP's best interest for that to actually happen. That's a long discussion, but it is not in our best interest. All right, thank you to all of our panelists and thanks to everybody for coming. OpenStack Cloud's all around the world.