 All right, hi, everybody. My name is Sean Madden. I work for Pissing Cloud Computing. I just here to do a demo and presentation on rapid VM cloning, as well as live migration. So here's the agenda. Just going to give you a brief overview of the technology that we use. I have a demo that I'm going to do, discuss the benefits of this technology. And then for the last five minutes, if you guys have questions, we can go over questions. And so just a little bit about talking about VMs in OpenStack. So usually what happens when we create a VM in OpenStack, we have to pick an image from an image store. That image has to get copied from wherever it's located on disk. And it needs to get copied to a local disk, which takes time. The image has to get booted. And that takes time to boot the OS, to boot the BIOS, to get everything up and running. And then we have to sit back and wait for several minutes, depend on the size of the VM, how much RAM you've provisioned to it, and stuff like that. So that's kind of not always what we want to do, especially if you have to provision a lot of VMs. And as you see this presentation, oh, got you. OK, so what I'm doing here is just showing an example of creating an instance. What we normally do is you pick an image that you're going to create it from. So in this case, we're doing a Windows Server instance. You pick an instance name for that. Then we pick a flavor, the size of the image. So I kind of prepped this ahead of time. So we're just going to pick a tiny image. And then what we do is we boot that image. The demo will kind of speed through this a little bit, because I don't have a lot of time. But the image gets spawned, picks an IP address, and then it boots. This particular Windows Server image, it took me this morning about three and a half minutes to boot to get it up and running to where it's in an active state. So it'll go a little faster than that now, just for the purposes of the demo. But I just want to give you the chance to see that, yeah, these images take a lot of time to spin up, and we don't normally want to wait for that. So it'll become active in a second. And I just didn't want to have you guys wait in the pain of dealing with the Windows Server image coming up. So it's active, and now we can actually spin this VM, get in there, get into Windows, and run applications. So that's the normal way that we spin up VMs. Hang on one second while it goes to the next part. So that's a default way. What we've done is we've implemented a technology called virtual memory streaming. And instead of having to spin up VMs from scratch every time, what we do is we pre-boot an image. So we initialize it, initialize the BIOS. The VM is already pre-booted, the OS is booted, and it's just waiting there ready to be streamed into action. So all the disk in memory state is already stored, ready to go, and all it needs is to grab a machine name, an IP, and we spawn it very quickly. And the way we do that is we create what's called a blessed image, which is this golden snapshot of an image. So starting from the VM that we just created, what you're going to see next is that we're going to create what's called a blessed image that will be able to spin up many VMs very quickly from. And this will be good for VDI solutions and stuff like that where you rapidly want to spin up VMs. So here, if we look at the dashboard, what's happening is you go into the dropdown menu and there's this bless link. All we do is create a name for this VM. I call it Piston Bless in this case, and we spin it up. So this is going to be the blessed image, which is the pre-booted image from which we're going to spin up many VMs. This blessed image took approximately the same time to create as the normal VM did. So for all intents and purposes, this took about three and a half minutes as well. So from creating the regular instance plus the blessed instance was probably about seven minutes total. I did this this morning in my hotel room. So it'll speed through. It'll go from the build state to active. So there, we're blessed. So what happens now is now that I have a blessed image, this is something that an admin might create in the morning and say, OK, we have our blessed image. Now I can go ahead and spin up hundreds or even thousands of VMs. So what we're going to look now is we take our blessed image, we create a name for it, and then I have a quota in here set to 10 instances. I can only spin up. So I've already spun up two. I have eight more I can do. So now we're going to launch. And what you'll see is you'll see eight instances come up on the screen. They're in the build state. They get their networking. You're going to see them grab an IP. And you're going to see them active very, very quickly. So they're in build and boom, they're active. So you have a blessed image already. We're spinning up VMs and now these VMs are ready to have worked on on them. So it's a very cool technology because now you don't have to wait. Yeah, you waited to make your blessed image. You waited to make your first image. But there, we have eight more images or instances that have been spun up very quickly. So let me just show you real quick. I can actually VNC into it just so that you guys make sure that I'm not lying about this. So we can VNC into it. We can get into the Windows server. And these things are up and running. So the virtual memory streaming technology is really cool, something that we're using at Piston to be able to, for those folks, it really need to spin up a lot of VMs very, very quickly. So let's let this move on. So there's virtual memory streaming technology is an extension to KVM. That's the hypervisor that we're using in our product right now. And the other aspect of this virtual memory streaming that we're doing is live migration. So we've actually added live migration piece in there as well that we're going to get into next. And with the live migration, that's good for us in a couple ways. Number one is we can take workloads that are on host and manually migrate them if we want to. We also have an update service that comes with our software that whenever updates happen, it's a rolling update. It happens live. So when there's an update, we migrate services off of host. Software gets updated. Hosts can get rebooted. And VMs can get migrated around as needed. And also, if hosts need to get powered down, any workloads that are running on those hosts get migrated off. So it's a very, very cool technology to be able to use. So I'm going to bring the dashboard back up and show a quick demo of how the migration works. And it's cool because all these features can happen either through the GUI or through the command line tool or through the APIs. All three ways from OpenStack. So if you look down here or look up there at what I'm doing, there is a migrate button. I can actually pick which host I want to migrate to. So I'm going from the .15 host to a .13. And you'll see that it shows up that the VM is migrating. I've actually sped this up a little bit for the purposes of the demo. I think when I did this in my room this morning, it took about three minutes to fully migrate over. And you'll see that it goes from migrating, storing on the .13 host. And then it does move over to the .15 to .13. So I sped through that. It's back active again. And so VMS helps us do migration, helps us spin up instances very, very quickly. Those are the two main reasons why we have VMS in our product. And we'll talk about the benefits very briefly about why these things are there. Let me see if I can maybe speed up a little bit. So here's the benefit. So with the virtual memory streaming, I showed you instances spin up very, very quickly. So there's really no waiting for VMs. You have a blessed image. You spin up hundreds or even thousands of VMs as long as your host can handle it. So it eliminates that boot storm in the morning. People come in, the whole company wants to spin up their VMs. You don't get that big boot storm of having to wait for all these VMs to boot with all the network traffic. VMS also does a lot of optimizations. And there's memory over subscription that happens. Sorry about that. Hang on one second. There's memory over subscription that happens. So when all these instances are coming up very quickly, we do memory over subscription. So then a lot of the memory that comes up is shared among those instances. So VMS helps us to about double the number of virtual machines that we can run per host, which is a really cool feature, too, because we have people that are interested in VDI want them to come up quickly, and they want to be able to maximize the number of VMs they have per host. And finally, I did mention the live migration piece, which was summarized, which helps us in our updates, helps us when servers power down to be able to migrate workloads, and just have that opportunity to migrate workloads as you see fit. So that's kind of the demo I wanted to give you guys. And I still have some time if anybody has any questions. We can do it all, man. We can do it all. But if you want to see any other demos, we do have a booth at the end over there, Pist and Cloud Computing. And thanks for coming. Right on.