 Okay, gentlemen and gentle ladies, I'm going to start with my presentation. This is going to be a nice story because the original title of this presentation was different. I'll explain you later why. But that doesn't mean we're going to tell a nice story about it. So I hope you enjoy this presentation and let's continue. Like I said, first of all, it's about how implementing services in a large telecom, what went good, how we continued. So we're talking about the partners, the transformation, what is learned and the ideal roles. And most important, like I said, it should be a story here about two people, two guys who met in early this year and did something together. Unfortunately, on Friday, we discovered that one of them was not allowed to speak. Yes, if you know why, I'm not going into the details, but it's not that he hates me or whatever. He's even in the room here, so it's okay. So who am I? My name is Frank Vines. I'm, to put it bluntly, I liked open source software. I'm in the business for 20 years with Linux. Last year, I went to OpenStack because I liked the idea about when you work together, you can create something unbelievable. And also, the last years when I'm in the sales engineering side of the business, I think if you want to sell something, it's about teaching. It's about educating your customer how you can use your products in the right way, how you can create value with the things that you deliver. Like I said, it's about explaining. It's not about preaching what's good or what's right. Decide together with it. But also, from a customer side, especially what I would call in this open stack world, you have to be open to learn. If you're able to learn, also be open, be critical, challenge it, give feedback to your suppliers, but also be a teacher in your organization. I can honestly tell you the implementations we did at Mirantis with OpenStack that were successful were the ones where we talked with the customer, we went together to implement a project where we created together value. I'll give you another example. Two weeks ago, no, two months ago, I received an RFP from an insurance company, 40 pages or something, and everything in that RFP was impossible almost because somebody had read a book about OpenStack and they said, I want 10,000 virtual machines running on SDN with software-defined storage in one cloud. That was a summary of it. I said, right, just do it, but not with me. When I want to do something, when I want to implement something, like I said, I've been in the Linux business, I've been the OpenStack business, and normally the model in the OpenStack, in the open source world, it's not licensing, it's subscriptions, and subscriptions has to be renewed every year. So if you create something that's not renewable after a year, you're stuck. So that's my mission. So like I said, I work for Mirantis. My mission is to build automated clouds, mission critical, and we are open to feedback because OpenStack, as everybody knows, is a rolling thing going on, and it's not one product that you ship in a box. Now, a large telecom, which I can't name. So in October, they decided they want to implement OpenStack in the data center and because they didn't have the eNAS experience, they decided to go with a partner and they selected the Mirantis for the anti-patient study. Like I said, this is a study what we did together. This is the journey to grow. Like I said, also what you see with OpenStack, it's what I call a disruptive software, it's a disruptive way because what's disruptive about OpenStack is, in the traditional data centers, you have networking, you have storage, you have server, you got operators. With OpenStack, you combine it to one situation. And not everybody loves that at the same time. You have to make sure, like I said, also when you want to implement OpenStack inside your organization, that you align also those organizations because otherwise it ain't be a success. So what was the idea about the transformation? Cloud infrastructure as a service based on OpenStack, software architecture designed for the cloud, the tooling and automation. The objective was to reduce the time to market in the life cycle. That's not what I would call big science, but that's the fact. At the moment, most of the customers, they have virtualization inside. But virtualization is what I would call a deluxe version of infrastructure. It's not an automating thing. So if you... Good afternoon, welcome. And also what you see in traditional infrastructure at the moment is a completely separation between, I would call the hardware, the service between test, acceptance and production. If you want to build a new agile infrastructure, one of the things is indeed is to make sure that you can deliver self-service. And with self-service, I mean not to the end customer, but internally to your project to the people. That when they need something, it's filled in. This is like in the real world, you know. If now you want to book a travel, what you do? You go to a website, you book the travel. You don't need the travel agency, the people sitting there like this slowing down the process. So also please use that self-service as one of the key features for new agile infrastructure. That also implements what I would say a more responsible responsibility about security but networking because it's also done automatically. The nice thing is if you implement OpenStar Clouds or Clouds in general in the right way, there's no difference anymore between the infrastructure for test environments and production environments. So the risk by moving from test, acceptance to production is much less than you go in the traditional environment. Also with the new application designs, sometimes they call it what I would call the cattle and the pets. Anybody heard about that? Cattle pets? Some people. It's one of the terms invented by Tim Bell, I think, from CERN, with a pet, I mean a virtual machine or a machine that needs to be kept alive. Whatever happens, it must be kept alive, you know? In the early days, you had nonstop tandem computers, make sure that everything was up and running, old thing for that only virtual machine. Now, when you design with new application, cloud applications, it's the size, the number of virtual machines which will create a high availability. Why is everybody opening my door? Is this a movie? Okay, we continue, like I said, oh, nice. It's like if you design new infrastructures, make sure that you scale in a horizontal way. Also, what I would say with OpenStack is it's better to have 500 virtual machines where 10% can die than five where only one can die. So that's the thing. Also, the nice thing about this is one of the nice things about OpenStack, and that's why I believe in OpenStack, it's the open APIs. And the nice thing is with OpenStack, because it's now really what I would call, I consider it myself, a standard in the industry at the moment, is that you can deliver APIs, programming interfaces to your internal customers, but also to your partners. That when they deliver software, when they deliver applications to you, that you are able to save on gentlemen, we are using those APIs from OpenStack. The high-style, the UNO, whatever. But at least you can say to them, this is the way you have to develop and deliver your software to. Also, the nice thing, if you go about virtualization and when you talk about it, it's in the early days of computers and service, you want to consolidate as much as servers as possible. Don't do it. Try even to get more modular instances, more things than creating big monsters. Also, one of the things with OpenStack is, and we did it with some other customers too, is gentlemen, you want to deliver as fast as possible. And because you have indeed, what I would call, no separation anymore between the test, acceptance and production environment, it's very easier to deploy things in production. Also, you can automate and make rules that for example, when you deploy it in production, you don't have to deploy it in production everything at the same time. You can create, for example, for these customers or these internal projects, we're going to keep the old version, but for 5% of the users, we're going to use the new version. And if everything works all right, we're going to grow the instance. That's the way how you deploy in a new way. Like I said, when you want to implement those things, you really need also to understand that also internally the organization has to transform. You have to indeed get those integrated teams, you have to know more silos, but you have to keep what I would say, even from smaller projects, you have to make it a complete entity. Like I was talking about, automation of tasks, testing development, exposed what is produced as a service. Also, one of the things that you see in the OpenStack world is when we speak as a service. Everything is going to be as a service. What does it mean is with traditional applications, you have an application and it requires some things, some kind of networking settings, a database, what I would call the add-ons to deploy and deliver that application. Now what you see in OpenStack environments is that those add-ons will be more and more delivered as a service, like I'm doing, a door as a service. But that's really the thing. You've got applications, I mean, users coming in, they need something, they open the door, they take it, and afterwards, with a good OpenStack environment, you also clean up the leftovers. That's also cloud automation, close the door. Also, more important, and that's also a key thing, is when you implement something new, when you deliver something new to a customer, to an implementation or whatever, is to make sure that the end user experience is good, it's better. So, for example, don't try to over-commit, create a big monster in the beginning, it's better to start with a small installation for a selected number of customers, but where they can make that project inside Atelko a nice success, so they can spread the good news. If you introduce what I would call sometimes a Franken-Stack in the company, it's going to fail. It's getting to get those guys with you. Also, what you see is indeed the physical architecture must be simplified. Also, more important to see and to do, is, like I said, the nice thing about OpenStack in Cloud is that you can start using commodity hardware, that you try to keep it as simple as possible. One of the things is you can't fix everything with anything, but you have to make sure that you start with an environment on commodity, which can be extended. Like I said, the NFVs, the services are key things in an OpenStack environment, because those are the add-ons. For example, Firewall as a service, the Groups, Load Balancing, L3, DNS, those are the things that in the early days with applications you had to deliver and to create manually inside the organization. From now on, you're going to deliver it by the Cloud. So, the telecom has chosen OpenStack. Why? Yeah, it was very easy. It's becoming the standard. I'll be honest, in the early days of OpenStack, last year, for example, there were still choices. Is there anybody in this room who doubts that OpenStack is going to become the standard? Gentleman gets up and leaves the room. Okay, nice. But, like I said, this is what it is. You've got the partners, you've got the ecosystem. Three weeks ago, I was in Barcelona at VMworld, and even there they were telling OpenStack OpenStack like in OpenStack, from VMworld, VMware's one. Okay, nice. We do it. So, that's no discussion anymore. So, what has the telecom learned the last 15 months? 50% of the value in what they did in the test phase is indeed coming from renovation. Like I said, a full automation is key of rapid delivery, improving procedures. If you can improve the way how you deliver, how you develop, how you create new services, how you go forward, this is improvement. And that's the most thing. So, OpenStack, it's not always about the money. I would say, and it's even what I've shown in service, it's about the flexibility. The renovation, where you also internally can rejuvenile your IT departments. 20% in the value in the network is the network fertilization, flexibility and scalability. I would even call that the most important difference between virtualization we have at the moment and cloud. It's the virtualization of the network. Because what we see more and more is that even networks in data centers get connected with other data centers and things like that. So, what we also see with global companies is make sure that how those networks are configured, also from a security standpoint of view. The thing is, in OpenStack at the moment, the change in networking components, how it grows as possible, is changing in a very rapidly speed. So, the nice thing is indeed, is also what I would say is, remark here, you have to reconsider almost, I wouldn't say not every six months, but every year that's coming, am I still using the right tools in OpenStack? The nice thing with OpenStack is also, for example, you can replace one part of OpenStack, for example, part of the networking of the storage, replace it by another component. We go from software-defined storage to a NetApp box or what, or we're going from an Arista switch to VXLan, those are components, can be swapped, can be used without interrupting your OpenStack cloud. But also, when you implement OpenStack, like I said, dearly, think about it. The changes, the decisions you have made six months ago can be different than what you would do now. Thirty percent of the value is a switch to open source. Like I said, I'm a PR advocate for open source. I've been in the Linux thing for 20 years. I hack my kernels personally. But for some people, it's still a change what open source is. Open source is you're going from a licensed model to a subscription, so you're paying for free software. I sell free software. And even my mother-in-law approves it. Yeah, what's free, even I tell, ketchup is free, you know? You can make your own ketchup, you can grow some tomatoes, you add some spices in, you cook it and you get ketchup. The nice thing with open source is the fact is, it's an open way to look at it. But for most of the people in the room, I suppose their business model is not making open source. It's using a product that delivers value to them and creates more value for them. However, because it's open source, multiple companies, multiple parties work together. And one of the things that I learned when I was in military school was when you work together in a group, the answers you create together in a group are better than when you do it alone. And that's the same with open source. If you see how many people at the moment are working together on open stack, are discussing it this week what they should do, that's not something that one company can do and learn. But in one company alone, you've got one vision. And when multiple vision clash, you get the creation, the value of the open source. Also, what I see is, for some people, and I truly, inside companies, developers inside companies like to contribute to open source. Not everybody, but most of them like it. Why? It gives them in what I would say, some presence. They are all alive. If you're just developing in a small room and you never, nobody knows what you're doing, that's, yeah, it's nice. Also, the fact is with open sources, there is no no. If there's a feature you need in your company and it's not really in open stack yet, there are two ways you can develop yourself or you can have it developed. But impossible is not yet there. Most of the time, like I said, we are top three contributors, we can recommend our customers what to do if they have to develop it. Maybe sometimes they have to wait because it's on the roadmap, but at least we can have a discussion about it. But that's also the nice thing about open source. Yeah. That was a conclusion about the telecom what they did about open stack. It's, like I said, they joined it. And the most important thing is what I say here and that's what I like about the learning here. After 12 months, we must challenge ourselves against the choice of the architect because we now we have the maturity to review our choices. Do you see the circle? And if you adapt to that circle, but if you're really saying that, okay, we're going to learn what we have done, we're going to improve it, we're going to do it again, we're going to improve and we're going to improve. So when I was talking about the insurance company who sent me an RFP of 40 pages for a five-year view, they had one idea. They want to implement it and there was no thing that we're going to review it. And that's not the way how open stack works. Bless you. Also, what they learned is major commercial players are now implementing enterprise-gride services. And that's like I said, you have to check out what your business model is, what you want to do yourself, what you want to have done for you, how we can do it. Also, what you see is indeed in the open source community, if you're a party, if you're a company who wants the influence, who wants to be on the blueprints, who wants to design even open stack, you can become an actor in the community. So yes, you can deliver code, you can upgrade. It's even more easy than most of the people think. And we from Marentis, we're also happy just to help you to do this. Other things that we also see around open stack is that you get complementary communities are emerging. They are relevant for the telecoms. What we see is, for example, we have now OpenNV, so network function virtualization. I got this gentleman in the room from NIV, which creates standards about it. I got a slide later also. And now, I think what my friend and I we learned the last 10 months, is indeed is when you're going to implement OpenStack, is that you clearly define on the beginning of this journey, because I call it a journey, it's not a project, it's a journey. What are the roles that you want to play on both parties? And I think, like I said, if you're the customer, you have to define the scope of the project, you have to define what's missing in your competencies, define the overall architecture of the project, prototype, learn, think. And we, as an integrator, we as the expert on OpenStack, as to advise the customer what to do or not. Like I said, if I got an RFP, which is fixed, line number one, ain't possible. I can't challenge you to say what's good and what not. We can't go into discussion what we think with our experience is a valuable solution or not. So that's what I do. So to be honest, if we can't do line one, it's a success or failure. The thing is indeed also for identifying the missing components, provide trainings, that's what you get best practices like that. Also what you do is some teams are thinking we're building on our own OpenStack, we're going to support it. It's your billing model, but at least that are the things what you do. So when you create and even you can shift one component to another site or whatever, but when you start a journey on OpenStack with an integrator, define those roles in the beginning, not during the project. This is the start of everything. Like I said, last slide on OpenNV. Like OpenNV, it's for telecoms. It drives closely with EtsyNV in the industry specifications. That's more technical side. But at least what we see also on that side is that we have corporate contributors to OpenNV. Companies like Ericsson, Huawei, Orange, all those guys, but also software vendors like Mirantis, and we work on that one. Gentlemen, gentlemen, ladies, this was my presentation. Any questions? Oh, thank you. Any questions, any ideas? Yeah, please. Hi, I'm Yuri Kutana and I've got a question. I was kind of intrigued by your intellectual challenge and the assertion that a group does produce actually a richer or more intellectual result. And I have been actually many times witness of the situation where a group was actually causing poor result just because it was a group and just it completely killed any possible intellectual outcome which could have been there if the work was a result of an individual or some high concentration of intellect. So I would actually doubt that as a general statement. But let me get me to the point. So let me present it as an area which kind of one may have different opinions on if I say it, just leave it more or less open. But this is what puzzles me. I mean, I began to look at all of these open stack things like three weeks ago and everything else was out there. And I found out that this great group effort is my opinion just trying to copy what Amazon is doing and is by far still lagging. All the things I hear here about, I found on Amazon and they were cool and great and working on the first side on the first try without all of these debates about some VNF and Yadi Yada and what Etsy does. And Etsy is, by the way, in my opinion, another example of group which has a great and long tradition of failures and ocean boiling. So trying to implement this ocean boiling activity doesn't almost sound to me like an example of how this group activity is going to lead to little result. So I don't mean to bash now open stack. It's just actually I'm attracted to buy it very much. That's why I'm here. But this is something which constantly keeps me puzzled. How come that actually it's still lagging behind Amazon? So you guys are probably longer in the business and can enlighten me why this gap is still here and what's going to happen to the gap in this great community. Okay, thank you. And sorry, I didn't. No, you can continue, but I'm going to give you an answer. Otherwise, I forget my brain is a little bit fried at the moment from the party yesterday. First of all, on the group thing, I did I did military school. And one of the things I learned is indeed how to create good groups because you can have groups with a poor result and good groups with good result. Rule number one is if you want to create a group which delivers better quality than individuals, create a hydrogen heterogene group, that's the thing. Otherwise, we can discuss it by beer about group things. The other thing is about yes, for some purposes, even maybe 80% of the purposes, Amazon is a good platform. And I'm not going to say that you have to go back from Amazon or whatever. The only thing I can say is at the moment, OpenStack is ready for some production workloads. Not everything. It's now about choosing your battles, see what's possible, what's not. If somebody says to me, if I'm frank, I got a VMware form of 2,000 servers, migrate me now because it costs too much money, I'm going to say next month it's going to cost you too much money because I ain't going to do it. But if somebody says I got developers who want to develop something or create new Cloud2 applications, they don't have to do it at Amazon because that also costs money, then you have a business case to run it on OpenStack. And that's what I say is in the beginning of the project, define your scope, define what's possible, define what it will deliver. I mean, of course, that's a good argument. You cannot run your own Cloud with Amazon. That's the point of control. I take it, sure. Gentlemen, gentlemen and gentle ladies, thank you very much.