 Welcome. I wanted to start off with just a little bit of a rah-rah. So this half of the room is open. This half of the room is stacked. Open. That's not very good. Come on. We'll try this again. Open. Open. Stack. Stack. Open. Stack. Oh, come on. You guys are still waking up. You might need a little bit of amped. Okay. I'm Cameron Cedar from SUSE. And I'm a systems engineer. And I'm going to talk today about multi-hypervisor environments. Before I do that, give you a little bit of intro about SUSE. SUSE has been in the business of open source for 21-plus years. We have 19,000-plus customers worldwide. We are a leading provider of open source software. And we have award-winning support to go along with that. So why would you use something like SUSE Cloud? I'm not going to go into SUSE Cloud in depth right now. But one thing that sets us apart, we do have 21-plus years of managing open source software. And so that plays a big role when it comes to the community of open stack. We know how to manage open source. We also have some of our products like SUSE Studio and SUSE Manager, which also complement the SUSE Cloud and open stack environment. And they work really well together. So some of the SUSE Cloud 3 highlights. This is our latest version that's out. Right now we're working on four for Icehouse. This is based on Havana. And it has now heat in there, which we can actually use. It's fully supported now. We have the telemetry stuff of Solometer. So we have some improvements. There's some cloud management, some new features. Full VMware support, which is awesome. Improved networking stack in block stores. We have things like the VMware, the EMC, the Cisco Nexus. SUSE Cloud 2 to SUSE Cloud 3 upgrade process. So that's really fundamental for an enterprise product. The addition of existing servers as cloud nodes. So you can go in, spin up some new servers and add them as cloud infrastructure. We also have delivered, just recently we made a big announcement for SUSE Cloud 3, high availability. I don't know if you saw the demo theater that was done by my colleague. Really awesome stuff. You can auto deploy a high availability solution throughout your cloud environment for compute infrastructure for the control plane. So what's the problem? Enterprises fear change. Cloud computing is something that's kind of fearful in an enterprise today. And there's several reasons why they fear that. There's various IT processes that they have to work around. They make large investments in their virtualized infrastructure. Whether it be VMware or some other type of virtualization platform. They invest in skilled employees. They get them trained on top of VMware. Makes it really hard to move if somebody's already been trained on one hypervisor platform. And maybe they get skilled in Linux, I hope. Or maybe they get skilled in some other area. Maybe it's Oracle. So they make a big investment there. So also lack of familiarity with open source hypervisors. So in my travels, I spend a lot of time with people that have a lot of VMware. We come up against this all the time. And I talk to them about how they can use workloads on KVM and on Zen as alternative hypervisors. Trying them out, testing those workloads out to see how well they'll run there. They don't believe me sometimes. But they try it out. They give it a whirl. Vast majority of VMware administrators are Windows users. So that's kind of a big hurdle. Getting them trained on Linux. That can be a big hurdle. That's a big investment, trying to get your staff retrained. Disaster recovery. So VMware has great disaster recovery tools. A lot of things that are built right in. You might not want to reinvent the wheel there and disrupt that process. So what do enterprises really need? It's all rainbows and unicorns, right? That's at least what we like to think, anyways. In a homogenous environment, that's not really true. These rainbows and unicorns don't exactly exist. So what do they really need? No data center is really homogenous. VMware has approximately 56% of the market. That's huge. Hyper-V is growing rapidly, although it's not necessarily on the tails of VMware. They're kind of growing in their own regard, in their own space. Cloud computing needs to be agnostic. We need to be able to plug in to multiple environments, be able to utilize current infrastructure, be able to take workloads and move them around from one infrastructure to another one, be able to have that brokering between various hypervisors. Today we have VMware, and these are the major players today. VMware, Windows Server, Hyper-V, the Zen project, and then KVM. These are the major players. Now there's some other things on the rise, but they're not mainstream yet. Mixed hypervisors support matters. Now why? Because you have workload optimization. There may be an application that runs really well on VMware. There may be an application that runs really well on KVM. I always suggest people that they try those workloads out on the various hypervisors, because you might find that you might be able to save some costs by running it on a KVM or Zen hypervisor rather than VMware and not paying into that VMware attacks. Cloud can simplify your control plane. You can schedule instances and workloads across the various hypervisors, and it gives you licensing flexibility on to be stuck in that VMware attacks. So how can I make it happen? With VMware, you're going to need a driver that's been added as of Grizzly. They've been doing a lot of work on that driver, and they really have expanded it quite a bit. Early on, it only supported a single cluster, which was kind of limiting, but now supports multiple clusters in Havana. Yeah, don't use Grizzly. So major upgrade in Havana. Compute node dedicated to communicating with vSphere. And so that can be easily done with OpenStack today. A direct connection to vSphere. There's also the NSX plug-in for Neutron, so if you are going down the road of doing your software-defined networking, you really want to deploy that. Clusters of clusters with VMware, so now multiple clusters are supported. And you're going to need a bridge interface defined on ESXi systems on your vSphere systems. Some of the limitations there, though, you have no IP tables. So where you were really used to that on KVM or Zen, you don't have that in the VMware infrastructure. You can't do security groups. You can't really lock that down, and so you have some limitation on that when you're running in VMware. No cinder support. Although in Icehouse, we do have the VMDK driver now in cinder, so that's kind of a nice thing. Take advantage of that. And then, of course, no live migrations. You have to rely upon the tools within VMware. So Microsoft Hyper-V. How do we do that? It's very usable as of Grizzly code. And you can set up some virtual switching. You'll need to do that within the Hyper-V environment. You have to enable the iSCSI initiator service within Hyper-V. So there's quite a few steps you've got to do here within the Hyper-V system. And you have to do the shared nothing live migration capability there in Hyper-V. And then install the Nova compute on your Hyper-V node. Some of the limitations, again, we don't have any IP tables, so we're limited again. No security groups. So if you want to go in and configure your users to have certain aspects of security within your OpenStack framework, that will not carry on over to your Hyper-V environment. And you must use RDP. So there's no way that you can use that console within Horizon. So the VLAN and routing is also supported on Hyper-V. When you use the Neutron Hyper-V agent, which is the ML2 driver, it gives you that Layer 2 functionality. So you're going to need that if that's the type of functionality you're going to want with VLANs. So Zen and KVM. This is kind of the de facto default around OpenStack using KVM or Zen. With SUSE, we give you the easy button. We make it easy for you to deploy, to consume, any hypervisor that you would like to use. Okay? So how do you do it? So this picture is a crowbar. We use the crowbar automation interface to auto-deploy all of our infrastructure for SUSE Cloud. Glance, when you are importing some of these other images for, let's say, Hyper-V or VMware, there's some things you need to look at. If you're importing an image, you need for VMware, you need to have the preallocated disk type. Make sure it's a flat image. Make sure it's ready for ESX. There may be some other images that may not actually work. So you need to pay attention to that. A nice utility from VMware is the VMware vDisk manager to do that conversion. So this is kind of a tool that you're going to need to use to convert that image into the right format that can be consumed in VMware outside of using OpenStack. And so this is kind of the end result. This is what it looks like inside of SUSE Cloud. You're going to have a Hyper-V, you're going to have KVM, you're going to have VMware all working together from OpenStack. So let's dig into that just a little bit deeper. I'm going to show you live some of this stuff. I've got everything running under VMware Workstation currently. And let me load up the SUSE Cloud dashboard horizon. And while that's loading up, we'll skip on over to crowbar nova configuration logged in here. And I have a couple of images in this environment. I have a VMware image and I have a KVM image. We can see those here. And they're automatically labeled VMX and KVM. We have a bit of a broker, I guess you could call it, SUSE Studio. We can build images in any format we really want, whether it be VMware or Hyper-V or something that's in Amazon EC2 Cloud even. But we can use that service to be able to auto populate our images in OpenStack. It's a great functionality. And you can see that here. I have VMX, I have KVM. And if I go to the basic project OpenStack, I can launch some of these here. And we'll just go ahead and launch a couple of these up. We'll start with the VMware one first. And I should have it pre-cached. It does take a little bit longer because it is, you know, it copies over the whole flat file for that image. So it just takes a little bit longer. So we'll go ahead and put that in there. And I'm just doing some very basic stuff here. And that's going to build and launch over on VMware. That process is going to take a little bit. So while that's doing that, why don't we go ahead and launch the one on a KVM hypervisor as well. And they're both spawning. That VMware one's going to take a little bit longer while that's going through that. Let's take a look at vCenter. So nice you can do this all on Linux now. It used to be you could only do vCenter from Windows. Not anymore. It's all logged in there. That'll take a little minute to load. And they're both running. I've got one running under VMware. I've got one running under KVM. And I could launch even more if I wanted to. The way that I'm making all this happen, using this broker that I was talking about, SUSE Studio, I have an image here that I've taken. I've created it. And I've selected all the software that I've wanted. I've gone through the software selection, created the image that I've wanted through this, gone through some basic configurations, set up DHCP, taken out the end user license agreement, personalized it a little bit. I can even add, you know, logo in there, whatever I like. And then also do some basic appliance modifications. This really plays into account when you're creating VMware images, the disk size for VMware. Right now it's set to two gigabytes. It will expand when you create a larger disk from OpenStack. And then also integrate it with SUSE Cloud. So we have the cloud init package there to be able to do the security settings and whatnot there for the GPG keys. And then we can also tie it in with SUSE Manager and get our full compliance in the cloud, patches and updates all ready to go for all the instances that we launch in the cloud. And then, of course, once we've set that image all up, we can build that image in various formats. And I can take that same image that I've launched in my private cloud and if I want to launch it in a public cloud, I could. I could launch that into Amazon EC2 or Windows Azure straight from the SUSE Studio Broker. Just by clicking that box, we can build an Amazon EC2 image. If I select VMware, I can go down here and click build additional and it's going to build me a VMware image. It takes about four minutes. It's pretty fast. Once it's done building, it's going to automatically show up in my OpenStack environment. We've got a listener running on the OpenStack side on our control node and it listens for that broker when it has a job that's finished. It automatically runs a glance command and brings that image right into the environment. So it'll be ready for consumption. Back over to VMware. Here's vCenter. Some of the basic configuration we've got in here. Let's go take a look at what we've got. Let's look at our hosts. More importantly, let's look at our clusters. We should have a SUSE cluster in here. Here's one of our hosts. Okay. So here's the basic networking stack. I do have two network interfaces in this particular environment. The vSwitch one has the necessary pieces. We've got the bridge interface all set up. If you look through the OpenStack documentation, there's some minimal settings that you have to set up in vCenter. Really easy to set up. Make sure you have that bridging interface and it will communicate really nice and easily. That's going to take a little bit. Does anybody have any questions? I showed you a lot of stuff. I went back and forth between several different things. SUSE Cloud. We started out with the Horizon dashboard. I've showed you the SUSE Studio, the image brokers so we can launch images into our OpenStack framework. I showed you launching instances on VMware and also KVM from the same dashboard. Really cool stuff. Yeah, question. I'm sorry. What exactly is an image broker? That's kind of my words for this. So it's basically providing the ability to create an image of an operating system so that you can actually take that image and import it easily into the OpenStack framework so you can launch it as an instance. That's really what it is. That's our product, SUSE Studio. You can go out to susestudio.com and play around with it there. It's free. Okay. So one question, which was kind of hard to see, but when you launched the different images into your different hypervisor environments, is the hypervisor kind of selected based on what the image format is? Because I didn't see a separate selection for image to hypervisor. It is. It is actually based on the image format that is used for glance when it imports that image. So in our listener that's listening from this image broker, he's actually recognizing the format and saying, okay, if it's VMware that's finished, he's extracting it, doing some modifications, and then he's importing it as a VMware image into OpenStack. So he's doing all that work behind the scenes for you. Okay. Hi. I was a little late, so I apologize if this was something you covered, but I noticed when you were booting the images, you just had one network there that you selected. If you're running a more complex scenario with a variety of neutron networks, number one, does it work? And number two, if it doesn't work, how exactly does the failure, if you select a network that is, say, a GRE overlay network? Okay. Good question. So it does work. And we're actually going to show that with some heat stacks on Thursday tomorrow. Same time. Well, actually, I think it's 11.30. It's a hands-on session. So if you want to see that working with some heat stacks, we'll be creating some alternate networks and launching those up as well. Okay. Thank you. Any other questions? It's pretty straightforward. SUSE is providing the easy button for you guys. It's really quick and easy to do this using Crowbar and SUSE Cloud. It's great stuff. The SUSE Cloud admin appliance is available. This admin appliance loads up Crowbar. It's got everything you need, all the software that you need to deploy an entire open stack cloud based on Havana right now. Here's the link. Go download it. It's really quick and easy. You'll have a cloud up and running about if you're as fast as I am a half an hour. Yes, question. That's correct. That's correct. So his question was, is it the ESX driver or is it the vCenter driver? It's the vCenter driver that we use. That's the one that's fully supported. The ESX driver was in partial not really written by VMware. So we don't use that one. VMware doesn't want us to use that one if we're going to get full support for it. So they want us to use vCenter. How is the HA handled? The HA DRS, I believe there's some requirements from the documentation. You have to make some modifications to it for it to work with open stack. So there are some requirements. So look out for that piece there with the DRS. There's some modifications there. Just minor modifications. And the open stack documentation lists out those modifications. Just two steps. Any other questions? We do have some geek-o handouts as you're leaving the room. We've got some cards with this link on it. So you know where to find the appliance so you can start deploying some SUSE Cloud. Thanks for coming, guys.