Hi everybody. We're here in the, I guess, the Menlo Park Sun living room. It's actually quite nice. There's a little kitchen maybe we'll show you later. But we're here with a special guest. Would you like to introduce yourself? I'm Steve Wilson. I'm the Vice President of XVM at Sun Microsystems. The sort of XVM group is a recent sort of grouping or brand, if you will, with Inside Sun. So what does it encompass exactly? Well, we started talking about this late last year, I think November 2007, and since then we built a portfolio of four products. Three of them are shipping now, and a fourth one will be soon. And it's really divided up into a couple different pieces. We have pieces specifically for the desktop side. We have things like the VirtualBox hypervisor, really popular. I think, I looked at the stats the other day, we're up to six point something million downloads of it. Right. So, you know, software developers, consumers, all sorts of people using that. And, you know, everybody from using Macs to PCs to running Solaris and all sorts of cool things like that. And also on the desktop side, we have solutions to scale up to doing large scale desktop consolidation. So say a business that wants to run, you know, today they might have a thousand desktops out in their office. And they want to get those out of the office because it's so expensive to maintain them. So you can virtualize those, consolidate them on a set of servers, and then manage them. And our VDI software can take care of all of the authentication about how users get to their sessions. Right. And they want to broadcast out either to existing PCs or dedicated thin clients like a Sunray, things like that. Yeah. I mean, how does that effort relate to the Sunray? Sun's been doing virtual desktop with Sunray for a long time. It's just, we didn't call it virtual desktop. It was great clients. And what's happened over the last couple of years is that as this technology has come around to do general purpose virtualization, we got out of the mode where you could only say,
2:00do Unix desktops with Sunray. So now, most of our customers using Sunray are using the VDI software to allow them to do virtualization with a full Windows XP or Vista desktop and be able to get access to that anywhere. You can be in Starbucks on your smartphone accessing your Windows desktop over the WAN now. So essentially, it's extending the Sunray idea to just any sort of... Any device and any OS you can now do. And the Sunray plays a key part in that because as people realize they want to get the same benefits they're getting from server virtualization on the client, what they're saying is, all right, I can virtualize all this, but if I still just have to have the same PCs sitting around the office, every time a hard disk fails, I got to send a tech out there to go deal with it. And they're sitting there burning 150 watts 24 hours a day, whether anybody's using them. So you have the ability to drop a Sunray down there. I think the current base model Sunray draws something like six watts. Right. It's like an LED or something. Yeah. And you can just leave that on 24 hours a day for 10 years. It never crashes. There's no moving parts. There's no state. So we've got the virtual box stuff and then the VDI stuff. Yeah. And then there's the two other components you were talking about. Yeah. And then you go over to the server side. So we have the Ops Center management tool that we've been shipping for a while. So there's XVM Ops Center. We started shipping that back in February. And that's really become Sun's mainline server management tool for how do you manage a data center. Everything from managing your server down to the, you know, how do I manage the firmware on the server? How do I bare metal provision the operating systems? How do I do patching and software lifecycle management and compliance reporting? Right. But to date, it's been very focused on the physical part of the data center. And so what we're going through right now is we're just finishing up the XVM server product.
4:02And XVM server is our dedicated bare metal hypervisor. So unlike virtual box that you install on top of an existing guest operating system, you install XVM server on the bare metal machine. Right. It sort of takes the place of the OS. And then you have the ability to do remote management of it through a browser user interface or an API or connect it to an Ops Center instance and manage it as part of your enterprise. Oh, right. So there's sort of like local command line-ish sort of management if you wanted to do that. But then you can also do it more remotely so you can sort of batch operations together. Yeah. So it's actually one of the things I'll show you a little bit later is we have a pretty rich browser graphical user interface built into every copy of XVM server. Okay. But it's dedicated to small deployments. Right. And it doesn't have the ability to manage the physical hardware and, you know, do patching and compliance reporting and all that kind of stuff. As you move up to Ops Center, you actually keep a lot of the same user interface, but now you add a lot more features to it to do all of that classic data center management stuff. Right. And you get one console that lets you manage the whole thing. So which sort of family of virtualization does the XVM stuff come from? I mean, we didn't want to reinvent the wheel for this. So it is, it's using some code from the Zen community underneath the covers to do the lowest level virtualization pieces. But it's also using a lot of technology from Open Solaris. And by taking some technology from the Open Solaris family, we actually get a lot of interesting benefits because Solaris has this long history in the enterprise. Some really interesting things that come out of that are the ability to leverage things like the Solaris fault management architecture. We've taken that and moved it in. And so, you know, when Sun started growing up Solaris in the nineties and running on big SMPs, one of the things you find when you run on a machine that has 64 CPUs is your odds
6:03of having a CPU failure have now gone up by a factor of 64. And if every time you had a CPU or a DIM hiccup, you know, your giant Oracle database crashed, that would be really bad. So Solaris had to learn to be resilient to things like that and could manage those faults, manage around them, isolate bad components and keep running. And so we're taking some of that same technology, putting it into XVM servers. So if you have a DIM that goes bad, if you're running Windows on that server, you get a blue screen of death. That's just the only, the only out. But with XVM server, if you were running Windows in a set of guests on top of XVM server, XVM server could isolate. Then you have the, the, the, the op center manages all of the XVM stuff. So like, I mean, you know, we were speaking at the server level, you have the, you know, you manage when hardware goes bad and things like that. So like, what's the more, what sort of management do you get when you have the operation center integrated into things? So op center, um, really, if you, if you think about the life cycle of a machine, be it sort of physical or virtual, but start with the physical one, cause it's sort of the most general case. You know, you start by ordering the machine, it shows up in a box, you unpack it, you put it in a rack and you table it up. Right. And that's the part you have to do yourself. Once you have that up, um, and sort of powered, you know, you got the power connected and the network connected, um, you can go to op center and do the rest of your management from there. Right. So you can remotely power the machine on, um, you can check its firmware level and make sure you have the latest firmware available for that system. Um, if you don't, you can update the firmware, then you can go ahead and put the OS out there with op center two, um, which is the one that will be shipping concurrently with XBM server. You'll be able to either put down a bare metal OS, like say, uh, you know, a Solaris or a Linux OS or deploy XBM server itself onto the bare metal. Okay. Right. So, you know, in effect deploy the hypervisor. So if you want to run a virtualized data center, you might get a hundred new machines.
8:05And the first thing you got to do is put a hypervisor on all of them. So instead of running around to each of them with a CD and installing it, we can, we can jumpstart all of those. So what's used to push that out? I mean, what, what kind of machine do you have to make sure you get that it's ready for that sort of thing? So, um, so in general, when you're going to run with op center, um, you may actually have one or even more than one management server. So you're going to have a management server in the center that we kind of call the controller. And, um, that controller is going to run the browser user interface and it's going to have a database that has the configurations of all the machines and sort of the, the source of truth for what's going on in the system. Um, but then, and, and for a, you know, small to medium scale deployment, you can just start managing servers there for a very large one, say thousands of servers. And we do have people using ops center with, you know, I think our largest customer right now is running 4,000 servers. Okay. Um, you'll create actually a second management tier, um, where you put a set of proxy servers and you can put those at different points on the network. Right. Um, so it, it does two serve two, it serves two purposes. One is it lets you scale out horizontally. So if you want to manage thousands of machines, um, you don't have to manage them as independent clusters, you can manage them as one central cluster, but still have the compute power to manage them. Right. And what we see is that, um, you know, in, in modern data centers, you have so many firewalls and things walling off different parts of the data center, but you still might want to manage that centrally and sort of the, the management protocols that go on to do these low level operations, like provisioning a server. These are not firewall friendly things there, they're, they're UDP protocols, they're broadcast there. And, um, what we see is a lot of customers who want to manage things centrally, um, with a lot of the existing management tools, you, you have to turn your firewalls into Swiss cheese to get, um, to get this to work. And so by putting the proxies out there on all your individual subnets, um, you can just
10:08have a very thin standard HTTP connection between, or HTTPS between the, the controller and the proxies. Right. And so, you know, again, to, to manage a, a small virtualized environment, there's actually a version of an ops center that just comes built into every copy of XVM server. And so there, there's nothing extra to install. As you get up to a medium sized deployment, you might have that one central management server that you've run up center and then run multiple servers from it. And then as you get up to very large, you can actually put that middle tier in and have the, the proxies and scale it all the way up. That makes sense. And so if someone wanted to use, uh, you know, XVM server on their existing data center, I mean, what would that scenario be like if they're not taking stuff out of boxes? Yeah, absolutely. So you, so what you're going to do is, you know, let's say you wanted to start just simply and try it out. You're just going to pick a server. Um, in the, in the simplest case, you're going to, um, you know, download an, uh, an ISO image from sun and, you know, say burn it to a DVD, um, and put it in the CD drive and, um, boot up the machine. You're going to go through a really simple install experience. Um, it's basically boots up as a, a live CD that says, you know, takes you through the install experience. If you've installed, um, the latest, um, open Solaris or even the latest Ubuntu, the, the first part of the install is exactly like that. It just sort of walks you through an install and be familiar to, to any sysadmin. Um, once you get through the basic install, then you're going to log in, um, with a, with a browser to configure the rest of the system and, um, it'll walk you through a pretty simple setup wizard to, you know, set up your networks and other things, your users. Um, and, and then you have a, um, management console right there in the browser that's going to allow you to create guests, monitor them, um, all that kind of stuff. And I assume there's sort of like a discovery process and things like that, that, that signed all the junk you've got out there. Yeah. So, so as you go up to, um, full scale op center.
12:09So let's say, um, you've got that data center full of stuff. Right. Um, exactly. The first thing you're going to start with is discovery, um, has a, a number of different discovery protocols that it can, um, work with. And if, if things are set up, you know, we have a lot of sun's new things are actually set up to be discoverable through things like bonjour and things like that. Um, and, uh, which is actually, that's the same protocol that Max used to discover printers and things like that. Um, and, um, for things that aren't discoverable that way, we have the ability to go in and discover them via SSH and IPMI and a whole bunch of different ways that you can scour the data center to find all the pieces. Right. And then once you have those, um, discovered, um, op center can push out, um, a little piece of agentry to each of them that, um, we can use inside the OS to do software life cycle management, patching, compliance reporting, all that kind of stuff. So last thing. So when I first encountered, uh, ops, ops center, it wasn't, um, it was just a general sort of managing your data center thing. Yeah. And a lot of the focus as we've been talking about is now around virtualization, which, you know, makes sense because that's what operation people care about nowadays. And so like, is there still the sort of more generalized focus on just everything you need to run your data center? Absolutely. I think it's a differentiator for us. If you compare to something like, you know, virtual center, which is just focused on managing that hypervisor layer. Yeah. Um, it's, it's up to you to figure out how do you manage the hardware underneath it and how do you manage the guests that are running inside it. So, um, you know, no matter how powerful that tool is, as long as it's in just that space, you're going to have to integrate a number of other tools around it. And I think we can provide a much more complete solution. Right. Given that we can do both pieces. So you still have the management into the, the guest, if you will, right into the, into the virtual. Yeah. And when you get into the, the compliance reporting stuff, for example, that's what it's all about is the ability to see into those guests and understand what's going on. Right. Well, that sounds good. We'll check out the demo. All right, cool.