 Hi, everyone. Thanks for coming. Last day of OpenStack. Glad you're here. And Mike Day and I will be talking about how some of these communities connect. So you're here to OpenStack conference. We're going to talk a little bit about a few of the communities, in particular about where KVM plays, where the Overt organization plays, and how that relates to OpenStack. So I'm Jean Staten Healy. I run Cross IBM Linux for the IBM Corporation Linux strategy and also Open Virtualization strategy. And Mike Day is our chief architect for Open Virtualization. He's our distinguished engineer on the topic. So he'll take the second half of the presentation. I'll take the first half of the presentation. We're all in a good place today because a survey that IBM does every two years and has done since 2004 talks about what's top of mind for CEOs. We also do a CIO study. But typically, during those surveys, the results have come back, saying that market factors were the most important. And this year, the most recent survey, it came back that technology factors were the most important. So I think that's one of the reasons that we see the explosion of an organization like OpenStack to over 8,600 members so quickly. So you guys are all here for that reason. Technology factors are obviously top of mind for you. You'll look on the charts back there and you'll see several communities that you're probably very familiar with. I just want to ask, how many of you are members of one of these communities or one of the open communities? Your company, what about your company? So we don't quite have 100%, but we almost have 100%. So that would be typical here. Sometimes I'm sure that you wonder how these things all connect to each other and how you can engage and how you can leverage the communities. And hopefully we'll talk a little bit more directly about that today. But what I want to start with is just what these communities can do. And there's some very well-established communities up there. OpenStack is a newer community, but some of the more well-established ones like the Linux Foundation, Apache, Eclipse, Hadoop. Very familiar to, I'm sure, all of you. These communities actually help speed innovation. They help companies to market. They help address those technology factors that are top of mind. So IBM has a very long commitment to open source, but not only open source, but also open standards. And we know that it's very important to be a contributor to the community, not just a user of the technology that's available. So when we talk about the benefits, we talk about the contribution, the growth, the enhancement, the pervasive support. So the enhancement, the pervasive support are very focused on what our clients are needing. And all of the things that the community does is that the standards that are out there, the speed of innovation, and the flexibility it offers our clients in the data center. That's really important. That comes from the community. So those of you who are not part of a community afterwards, I would love to talk with you to just understand what you're thinking is around it if you're thinking about joining a community, but maybe why you haven't joined one. And I'd like to address that with you. One of the things that we are focused on in the IBM corporation right now is open virtualization. So we know virtualization is the on ramp to the cloud. And we also know that OpenStack is a focus cloud. Virtualization can't really talk about one without the other. And our focus on open virtualization is KVM. We chose that. We've been working on that very steadily for the past five or six years. That's been the one that we've chosen to focus on. Some of the statistics from the analysts, I'll show you in a second, we'll talk about the pervasiveness of open virtualization, especially KVM. You'll see that in OpenStack. Some of them are saying it's 95%. That's IDC. Whatever it is, it's a large majority that KVM is being used on. But one of the things that's really important is that as you go and build a cloud infrastructure, we're looking at building that infrastructure with open software, open services, and we've made an announcement and commitment to that. So that's our Platinum membership in the OpenStack organization. But I also wanna say that we look at three real priorities up here when we talk about Open, and especially as it relates to the cloud, that it needs to be user-driven. So one of the organizations that we have what actually helped found was the cloud customer services organization. It was really built on giving recommendations from clients and customers to build clouds and make it a little bit more accessible. So it's more user-driven. So that's a really important aspect in terms of these open source communities. So a cloud support, a customer council. The second one is that it has to be proven in practice. It can't just be theory. So one of the other aspects of reasons to join OpenStack is that it's, well, you've got 8,600 plus members. So somebody's doing something right. And also the technologies of the openness and the community contribution just really enable it to be more ready for the enterprise, the code being ready for the enterprise. And this is really important today. It may be more important than ever because the rate of innovation that is expected from companies is accelerating to the point that if you aren't on the cutting edge, you really don't have the differentiation that you need. So open communities are allowing for that differentiation to also stand out with companies. So that's the second one. The third one is that you need, when you're talking about a cloud, you need a standard architecture. And IBM contributed a cloud computing reference architecture through the open group to, you know, it was really kind of the basis for some of the cloud architecture. So we're quite committed to these open communities. We believe in these communities. And we're looking forward also to working with you, a very large community such as OpenStack, continuing to work as a major contributor and also a user in our products and services. So as I said, this talk will kind of tie together for you where we are focused on virtualization and with OpenStack. KVM is, as I said, is our choice for the open virtualization hypervisor. It's got some great performance benchmarks. The TPCC benchmarks just came out not that long ago as the highest ever on the x86 machine. We also have the top seven benchmarks for specvert benchmarks that are published. You can go look at these out on the web. So the performance is there on an open hypervisor. The scalability is there. The security is also there, which is important. It takes from SE Linux, has EO4 plus evaluation, common criteria evaluation certification. So that's also critical to have an enterprise ready hypervisor. So KVM is enterprise ready, it's out in the enterprise, it's deployed. We're finding that those who have proprietary hypervisors are also looking at a second hypervisor. So the analysts support this as many as two thirds, according to some analysts firms, have KVM as a second hypervisor. It's not necessarily primary yet in production, but I think there's a lot more experimentation going on right now with workloads, new workloads, white space. The rate of virtualization in China, for example, is a lot of white space, but the rate of virtualization and adopting it is significant. I said that I would show you a couple of quotes from analysts that talk about percentage of KVM usage. So you see the IDC one there, it's saying the 95%, is their estimate, and then Gartner is also talking about the high levels of open virtualization. Another organization or community that we formed to really gain more market awareness of KVM. We formed it in May 2011, so it'll be two years. We announced it at OSBC a couple of years ago. We'll also be back at OSBC, so if any of you guys are at OSBC at the end of the month, it would be great to see you, and please say hey, I was an open stack. But we formed that to really gain more market awareness of KVM. And I just wanna ask you guys, how many of you are at least experimenting with KVM in your data center? Okay, I would expect that from this crowd. So that's great, very high percentage. Not all of you, but very, very high percentage. Very few hands were not raised. Anyone have KVM as primary? That actually surprises me. There's several here who have KVM as primary. So definitely would be interested in talking with you afterwards, and also if you don't have it primary, we would like to understand what it would take to make it primary, so what you think the inhibitors are, and I think this would be great feedback for the community too. But we started the OVA, the Open Virtualization Alliance with about seven members. It grew to 250 quickly. I think what that really said is that people wanted to learn more, and also we've now changed it so that it's a place where if you have, if you're a member company, you can actually, I think I hit two of them, you can actually post blogs there, you can put your white papers up there, you can put your customer references up there. So it's really meant to be a community place where people can say, hey, this is what I'm doing with it. Here are some real live customer examples. That's always important to gain adoption. So we want to accelerate adoption of KVM through this particular open community. Now, a sister community to that that was formed not too long afterwards is overt, and Mike is going to talk more specifically about overt and how that relates again to OpenStack KVM, but overt was really formed to have an open stack around virtualization, starting with the management layer. So it's formed with several of the same companies that you'll see, seven companies up there, you'll see that we're on the open virtualization alliance. So it's more the technical side of it, whereas the open virtualization alliance is more of the awareness, you can get your stories out there. And we like to leverage the community for that. So we do a lot around the stories of the members of the community to do webcasts, if you wanna do a webcast. If you want to publish something out there, you're welcome to do that. So I'm going to turn this over to Mike who's going to talk and do a little bit deeper dive into how overt itself is leveraging KVM and leveraging OpenStack. Mike? Thank you, Gene. All right. All right, I think you can hear me now. Good. Okay, thank you for your time. It's a privilege to speak with you this afternoon. And it's been quite a journey in just a couple of years with Overt. And the process of seeing OpenStack get a lot of mindshare and it's been really great. OpenStack is a great community and it's been good for Overt as well. And a lot of you may not know very much about Overt. Some of you know a lot about Overt. And by the end of this presentation, I think you'll probably know a little bit more. You might learn a few things about what we've gone through as community members and community leaders and KVM developers. So the first thing I wanna talk about a little bit is technology that does projects that do the same thing or communities that do the same type of technology. Is it a waste of energy to have two projects that do the same kind of thing? What if we're talking about file systems, for example? How many file systems are in the Linux kernel? We have more than one clustered file system, for example. And for example, what, I heard people say that we don't need butter FS because we've already got the X3 file system and then later they said we don't need the X4 file system because we've got butter FS. There's always examples of things like that. But the fact is that if people need something and if developers need something, they're going to do that development and work on that technology because it's providing value. And so technologies might appear to be redundant or overlapping, but if they're providing a unique value, they'll get done and people will support them. So with Overt, Overt provides virtualization management, VM life cycle deployment, a VM image repository, VM storage services, sound familiar? That's a lot of what OpenStack does. So we actually had more than one person within IBM who in OpenStack Mania took hold within our corporation tell us to stop working on Overt and then not only that, but convince Red Hat to stop working on Overt and RevM at the same time. So that was kind of a difficult thing, but it turned out that we had to justify our existence to not only ourselves, but to the rest of the corporation and we couldn't just say, well, this is different, we need to work on it. Had to have a good explanation. So another way to look at it is what if open source projects were guitars? Any guitar players? Okay, so I hope you don't get upset with this or I hope you enjoy it. So there's more than one type of guitar there's many different types. This is an acoustic guitar and it's got a hollow body and kind of a nice warm tone. People can play it in all kinds of different ways. And originally when Fender came out with their solid guitar, a lot of people were just dismissing it as real crude instrument. This isn't the original one that they came out with but it found a lot of people, a lot of people found it useful and this particular one is different in some specific ways, it's very popular and it has a real bright tone. It really comes through a lot of the noise. This is the Les Paul guitar. It has two double coil pickups, a real fat tone. A lot of people like it. A lot of classic rock and roll you'll hear has this guitar tone and if you put them all together, they all sound better together and if you go back and look at a lot of the best music people my age listen to, you'll notice that they try to put different guitars together in the same band that they get guitar players that they play different guitars together. They try not to pair the same guitars because when they put them together they just don't sound as good. And I like to think that Overt and OpenStack kind of fit that analogy because they're really complimentary technologies. They solve similar problems in different ways. So to back that up, I'm gonna talk a little bit about the history of our OpenStack and Overt development from the perspective of the Linux group and IBM Linux Technology Center and IBM is a big company of course and we're just one part of IBM. And then what happened recently and how we reacted to the emergence of OpenStack as a major force in the industry and a major strategic goal and major strategic technology for IBM and then a few closing thoughts about how technology changes can affect open source projects for good or bad and how open source projects should react to technology changes. So OpenStack and Overt, in the LTC we started working with OpenStack back in 2010 we had a few engineers who were working on the KVM hypervisor to go down to the OpenStack Summit in San Antonio. They were very impressed and they wanted us to start working on OpenStack and use that as our management infrastructure. And it didn't work out very well. One of the things was that some of our executives didn't like the governance model but the technology was kind of neat. Engineers have always, every engineer that's ever looked, excuse me, at OpenStack has just really loved it and loved working on it. These guys weren't the only guys within IBM. I think there were pockets of engineers within IBM all over the company about the same time looking at it. In August of 2011, we worked with Red Hat and we recruited some other companies with Red Hat to start the Overt community. Now Overt is based on an open source management infrastructure for KVM and it's the upstream source code for the RevM management infrastructure. It's a significant product. So it's aimed at the vSphere type of use case, use model and it's got a lot of functionality. So this was a major undertaking and a major community that we stood up with Red Hat. So it's not something that we could just walk away from once we got this started and Red Hat of course has a business unit around this community, around the upstream community and the downstream product support model and everything. And then in 2012, IBM, Red Hat and many companies formed the OpenStack Foundation and we at IBM and also at Red Hat just literally went nuts over OpenStack and every product group has an OpenStack strategy and I say that, that's good. It's good but there might not be the right level of understanding about the open source process and how you can use OpenStack in your products and what that really means. So there's all over the map in terms of how OpenStack can affect your development process and how you can use OpenStack in your products and one of the things that happened to us in Overt is we started getting pressure to stop working on Overt and so that's where we're back to the comment I had in the earlier slide. So to justify our existence, we needed to go and really show what the focus of these projects are. Why Overt as a project needs to continue existing if OpenStack is strategic, which it is, which I support 100%. So Overt's charter is to exploit the KVMI provider and at the time we were having difficulty with management infrastructure for KVM. We were having a hard time. We were developing new features for KVM faster than Lib Overt could expose them and there was no easy way for users to take advantage of new KVM features. Lib Overt is a multi-hypervisor management library. It was frustrating and so at the time we decided that the Overt project would be a KVM first management infrastructure. It's going to exploit KVM. It's going to expose every KVM feature and we're going to use it so that KVM is always the best, you know, has the most features and it's got the management interfaces immediately when a new feature is ready. So that contrasts with OpenStack, which is a multi-hypervisor environment and it uses abstraction so that it really doesn't want to know too much about the hypervisor and it uses that abstraction of the virtual resources and the hypervisor to achieve massive scalability. Now Overt is not that scalable and because of the detail that it knows about and will continue to know about, it's difficult for Overt to achieve the level of scalability that OpenStack does. OpenStack works best with a few types of hosts and VMs deployed on a massive scale and this has been confirmed or reconfirmed to me at least when I've been listening to some of the discussions, the development discussions this week going on. I think it's a founding principle of OpenStack to delegate a lot of the detailed management or the hypervisor specific management when you get into an area that is specific to a hypervisor to let the hypervisor handle that. So that is for KVM, that is the job of Overt and so that's why Overt exists. Now yes, there is duplication. So what do we do about that duplication? That's an opportunity for the Overt community to integrate and to add value to OpenStack and so the following few slides are a few ways to suggest how we may be able to do that as a community. One thing that I didn't mention that underlies everything that we've been doing since 2005 when we started working on Zen, 2007 when we started working on KVM is we need to have an open hypervisor and open source management APIs and open published code to manage the hypervisor. Now I've tried several times to download the VMWare management API and every time I get to the download screen it sends me to a customer login. So at that point I just closed the tab and we don't ever want that to happen for KVM. If you want to manage KVM and you want to use the APIs you can get the published APIs from the web without signing a license agreement or becoming a customer of anybody and it's always going to be like that for as long as we're working on it because that's why we're working on KVM and we're working on Overt. So at the most basic level, layering and redundancy start off with OpenStack of course doesn't actually create a virtual machine. It invokes a hypervisor specific library to create a virtual machine. So that's layering. Layering is technically not redundancy and so this is what it looks like if you want to generalize it for KVM and for Zen and for PowerVM as well. So that case is pretty clear. Now Overt has distributed block storage services that have traditionally been pretty advanced with VDSM that's a distributed cluster file system that's good for work group size clusters. It's not enterprise level storage but it's a really neat cluster where each host provides its block device, integrates it into the cluster. It gives you shared storage resource. VDSM does that. Now with GlusterFS you have some pretty sophisticated shared storage that's provided by Overt and we've worked on some translators that give you storage offload and some other things. So by supporting the Cinder API for both block and LVM on the Overt node we can provide Cinder storage services to OpenStack and at the same time those storage resources are available natively to other Overt nodes. So that's an example of something we could do to add value. Remember if we don't add value we won't exist. So these are some things that we might be able to add and this is not the only way that we could add high-speed snapshots for example to OpenStack. There are other technologies obviously that can do that like filers and things like that but this is valuable. So image scheduling and placement for example. There's a lot of information available from KVM and from the Linux kernel that is useful for filter schedulers. So we could provide a set of filter schedulers that provide information that's important for KVM that probably doesn't apply to other hypervisors but could be very important for placing VMs and this is an example of a way to add value. It might be important to have information from Cilometer that only applies to filter schedulers for KVM and I'm not sure how that will work or if it'll even be standardized but it's something to explore and it's an example of the way that as over we should look at this. There's a way to live in the OpenStack world. So for Glance, it's pretty simple. Over it has an image repository and Glance is an image repository service. Glance is pretty simple right now. It serves some block images and it was an interesting discussion on Tuesday afternoon about should there be a canonical format for image services and in OpenStack image files or not and I think right now it's going to stay as a block image for a while but that's okay because it's an opportunity for Overt to provide images as block files to OpenStack via the Glance API and also to consume images and bring them into the Overt world using this type of model and it provides a benefit to both sides if we could do that. So these are the kinds of things to look at and I'm pretty sure people are already looking at doing this kind of work and then finally within an OpenStack cloud you're going to have workloads that are different from those workloads that you deploy en masse. They might be high performance. You might configure them with SRIOV hardware. They might be partitioned instead of sharing resources. You might need to manage some specific parameters that OpenStack doesn't know about. They might be HPC. They might be extra secure and they'll exist as clusters of Overt managed nodes within the OpenStack data center. So a way for Overt server to share information with OpenStack so that you can use OpenStack to manage these clusters or islands within the data center. So to wind it up, OpenStack communities will only exist as long as they provide value to users and developers. For Overt, our original goals were to create an ecosystem for KVM ISVs to also provide a unified management infrastructure for KVM and to expose and support KVM features in a timely manner. Now it's probably true that OpenStack has fulfilled that first goal to create an ecosystem. But Overt is really continuing to fulfill the infrastructure and feature goals so it's continuing to provide value for us. And there are new opportunities for Overt to provide value to OpenStack data centers. So I think that's an object lesson to OpenStack communities that when new technologies come along that you could perceive as a threat, they also might be opportunities. And if you are using the right technologies and you are using the right license and you have the right culture among your developers, then you can react and persist as a community. So hopefully that's what we'll be able to do with Overt. I know it's important to IBM. I know it's important to Red Hat and the other companies who are working on it. We can't manage KVM the way it needs to be managed without Overt. We can't deploy KVM on the scale as required by our large customers without OpenStack. So that's the reality. Thank you.