Remarks by Debra Chrapaty, Corporate Vice President, Global Foundation Services
Microsoft Management Summit
May 1, 2008
ANNOUNCER: Ladies and gentlemen, please welcome Debra Chrapaty, Corporate Vice President, Global Foundation Services, Microsoft Corporation. (Applause.)
DEBRA CHRAPATY: Good morning. Hey, before I get started on my speech, I do have to comment for the 2 percent of the women in the room about what just happened up here. I think it’s the morning for women, ladies. (Cheers, applause.) So, a great job on that.
So, my name is Debra Chrapaty. I’m the corporate vice president of a group called Global Foundation Services. And what my team does is we essentially run the infrastructure for Microsoft. By that I mean the datacenters, networks, servers, and, very relevant for this group, the manageability.
So, I know that you heard a fantastic speech from Bob on day one about the Dynamic IT for datacenters. I know Brad absolutely rocked it on the second day in his presentation on Dynamic IT for the desktop. And what we thought that we would do this morning is kind of present a slightly different perspective.
What I wanted to do is take the next hour and share a perspective on how Microsoft runs its datacenters and infrastructure, sort of an inside view for you to set some context on how we actually do it in side the company.
So, let me just kind of give you a perspective on Microsoft. I know you’ve heard about our work in software plus services. If you’re keeping taps — and, of course, I’m sure you are — you’re hearing Ray Ozzie and other folks out in the industry talking about our initiatives around services. It’s touched every part of the company. Whether it’s our Live initiative for the consumer marketplace with Windows Live and search and Hotmail and Messenger, which touches hundreds of millions of people; in our enterprise group with Office Live and Exchange Live; and even in our enterprise and devices group Xbox Live, there’s a huge services initiative.
So, within my group people think about services, and they think about the cloud. So, we talk a lot about the Internet cloud, and I’m sure you do in your world. Where is the data stored? It’s stored in the cloud, right? What am I building? I’m building infrastructure for the cloud.
I talked to my grandmother a couple of weeks ago, and she said, “Debra, what’s the cloud? What is this cloud?” Gosh, she kind of has a picture of the cloud in her mind that looks like this, and probably your kids do or you cousins do. They think it’s a cloud.
And I hate to ruin your breakfast, but actually that’s the cloud, right? The cloud is the datacenters, the networks, the infrastructure, System Center, the stuff that manages it and the stuff that runs it, the stuff that actually enables us to deliver to our customer base.
So, within my group, Global Foundation Services, we essentially run all these things on this wonderful foundation for the company, and I wanted to give you a perspective on the size and scale of Microsoft.
Sometimes even I find it just overwhelming in terms of its size and scope. It’s one of the largest infrastructures in the world, and we’re running by, and these may not be numbers that are familiar to many of you when I talk in terms of in Live Search, for example, over 2 billion queries per month. We’re talking billions. Our index is including over 20 billion documents and 400 million images, and over 7 million instant answers. These are pretty significant numbers. Within MSN we talk about 550 million users.
So, for example, if I get a call in the middle of the night, God forbid, and it does happen rarely but it does happen, that we have some kind of service interruption on Hotmail, and I say to my team, well, how many people have been impacted, and I get a number like, oh, 220 million people. Imagine getting a call like that if you’re running a service for your company, and you get a call, oh, 300 million — probably you want to get out of bed, right? You want to get up and kind of address the issue when 300 million people are being impacted.
So, MSN, for example, 42 markets, 21 languages, 10 billion page views per month — not million, billion; and Live ID, things like over a billion people authenticating on our Microsoft services; Hotmail 350 million users; Messenger 8.2 billion messages per day, and I can go on and on and on with what’s up on the screen.
Think about the complexity of that. Think about the manageability challenges in an environment that’s that big, that’s handling that much volume; pretty significant challenges that we face. And I’ll get into it later about how we run our datacenters and the manageability tools we use, but we’re looking at System Center to help us to work through this complexity, to capture these performance numbers and bring them back into our environment, manage it, and better because we can measure it, focus on it, and manage it more effectively.
So, let me give you a perspective in terms of the infrastructure itself. So, I’ve been with Microsoft a little over five years. When you look at the numbers here in terms of server growth, in those five years, when I came to Microsoft, I thought that the environment was pretty big, certainly large than anything I had ever run in my career. Five years later, we’ve got 15 times the number of servers in our environment. Think about that, 15 times the number of servers.
I’ll give you a perspective: We’re adding about 10,000 servers a month, by the way. How’s that for hyper growth? A month. Our egress number is nine times larger than they were when I first came to the company. Three times the number of datacenters, and we had a lot of datacenters five years ago. And power, of course, is something that I’ll highlight in my presentation because it’s something that’s becoming so essential to think about, so essential to be able to measure and to manage, because it’s becoming a limited resource and it’s very expensive, you know, 15 times the power, so a pretty significant number.
And here’s the good news for me in terms of growth and job stability. Our growth is expected to continue and to increase in the next five years, doubling at least the server growth in the next year.
So, again you think about that, and what resonates for me, and it’s top on my list, is how you manage this. What do you do with this kind of complexity? Well, System Center, of course.
So, building and running at scale. What I want to do is just walk you through a couple of these components, and just give you some highlights, and then we’ll do a deeper dive on the manageability piece.
But, you know, if people ask me, you know, what do I focus on most and what’s the biggest challenge in my role at Microsoft, it’s building and running at scale.
So, what I want to talk a little bit about are datacenters, a little bit about what we’re seeing in our server environment, touching on network and trust, and then manageability.
Here are some picks, and they’re some pretty fun picks.
So, for those of you again that are kind of reading about what we’re doing in the press, on the far right is the San Antonio datacenter that’s going to open in July. Down on the bottom is Chicago. This datacenter we’ve broken ground on. This will be the largest datacenter in the world, pretty much, as far as I know, certainly in the United States. It will be over 100 megawatts of power, so a pretty big facility, if you guys out there and ladies track in power numbers.
The bottom is a Dublin rendering. It’s a picture. So, if you do out to Dublin and you look for that facility, you won’t find it yet, but it’s going to look like that when it’s done. There’s a picture of me with a Taoiseach standing in front of this rendering, and then like a couple of weeks later he was gone, if you follow Ireland politics, and that was a little surprising. So, I have that picture up on eBay if you want it, because I don’t want it anymore.
And then that is Quincy, Washington, and that’s actually a picture of a groundbreaking when we broke ground in Quincy, and if you have good vision, that’s me in the backhoe. I tried to use the shovel, it didn’t work. So, they said, “Debra, get up in the backhoe.” So, if you don’t think you get to have fun at Microsoft as a senior executive, check out that picture and you can see that you do. I recommend you don’t wear white pants, though, which I did — the last time. I didn’t do that in San Antonio.
But let’s talk for a minute about datacenters. How many people here build datacenters or are involved in any kind of datacenter growth or build?
So, I think we all realize that this is becoming a pretty scarce resource. We’ve come to realize that at Microsoft, you know, as we deploy services and building that cloud, managing that cloud, it’s a datacenter. Ten years ago, let’s face it, they were being sold at 10 cents on the dollar. You could lease at 10 cents on the dollar. With Microsoft moving to services — oh, and by the way, pretty much every other company you look at moving to more of a services initiative, datacenter space is becoming premium. It’s becoming a defining and differentiating resource for companies like Microsoft and others.
So, when we think about — and I always use an analogy around like FedEx, you know, what made FedEx the most successful distribution company in the world? Was it that they had the best planes or the coolest logo or whatever? No, it’s because they put strategic landing fields in key distribution areas so they could get in and out faster than everybody else. The same thing is true when we think about datacenters.
So, we think about, for example, location, which I’ll talk about, but first of all smart growth. These things are expensive. I mean, a small one is 575,000 square feet. It’s probably about a half a billion dollars to build one from the ground up. So, these are pretty big numbers. They stay on your books for 15 years. You can imagine that Steve wants me to be really thoughtful about when we bring these things online and they start to power our services engine. So, we talk about smart growth. I always say to my team, and we use this analogy, that every kilowatt counts.
When we’re out there talking about companies, competitive companies, some of them, who I won’t name because I don’t like to reinforce their brand, I never say the name in public, but some of these companies — I use the word “G” for example but I never say the name, because it reinforces the brand — they are just taking out space, taking out space. We’re not. We’re really trying to be thoughtful in buying land, provisioning the shell, and then bringing it live in tandem with new services. So, smart growth, every kilowatt counts.
Then location, location, location. When you think about back to that FedEx analogy, we need to be in critical locations where we can have great performance in bringing data in and out quickly.
So, if you see the picture of this site selection heat map, what we did is we created 24 criteria, weighted criteria of places where we thought datacenters should be optimally placed. And these criteria are a lot of the things that you probably can guess, you know, network bandwidth, power availability, clean power, and sustainable power. I can’t tell you a lot because they’re sort of our secret sauce, and then you might show up like we did, but we showed up in Quincy, we flew in, we picked the location because of our heat map. When we were leaving, then we saw Yahoo! show up, and then we saw Google show up, and we saw Ashe show up and Intel show up. I think that they’re following me around. No, I’m kidding, but they showed up right after us. And we’ve seen that our heat map is actually starting to set trends in local communities around where to put these facilities. It’s such a critical decision, huge impact on your costs.
Innovating for efficiency and sustainability, and I’ll talk about that in my next slide for just a minute, because I think preserving our world and being really thoughtful is key to Microsoft’s overall strategy, and I want to share a little bit about our thinking there.
And then operational excellence, back to the numbers I showed before. When you think about this complexity, to me if we can’t manage it, we can’t measure it, we can’t run it. So, making sure that we have the right tools in place is absolutely critical.
So, part of operational excellence is the tooling of these hundreds of thousands and maybe even millions in my case of servers looking down the road.
So, let me talk about sustainability for a minute. Obviously these big datacenters have an environment impact, and they also have huge costs. The power costs in the datacenter, for example, the power and cooling, if you look in the datacenter, power is about 30 percent of the cost, and cooling is about 40 percent of your power costs. So, efficiency in these numbers is really critical to success in your overall sort of delivery of your services to market in terms of contribution margin, operating costs, et cetera, things that, of course, are part of the balance of decisions that we make.
So, when you think about building a datacenter from the ground up, certain things that we’ve been putting in place which we’ve found to be very, very helpful in power management are things like outside air cooling. Right, again, if 40 percent of your cost is in cooling, using outside air to cool the datacenter, critical as part of your ground up design in terms of sustainability, hydroelectric power, we also use external chillers that then cool the datacenter, right, through externally chilled water, making sure you use gray water as opposed to potable water. We all know that water is becoming a critical and scarce resource, unfortunately, particularly in certain parts of the world. We use reflective emissive paints so that the heat doesn’t penetrate the datacenter, just keeping it generally cooler.
We’re also doing some other innovations. You know, when you think about the servers, you can just design servers that use less power, and so we’re doing that. We’re working on our own server design, our own motherboard design, chip design.
We’re also working with Bob and Brad on not just the System Center to run these servers, but also changes in the operating system so that we can really maximize the power usage within these datacenters, because it’s such big numbers.
I always say to my team that a really tiny number times a really big number is still a really big number. So, if we can just whittle away at those power costs, big savings for Microsoft absolutely on our radar screen, while making sure that we have a focus on sustainability.
You know, we’re a part of a number of industry groups. SteveB was out at CeBIT a couple of — I guess it was last month, talking about our innovations in sustainability and green. We’re part of Green Grid. We’re part of Climate Savers. So, just keep in mind as we build and run things at Microsoft, we’re really focused not just on building them but doing them right and doing them with a real commitment to sustainability.
Server trends: So, you know, from a hardware perspective, and you guys are seeing it, too, you know, it used to be a single core and then it’s dual core and now it’s multi-core. I just was sitting down and looking at a session with AMD and they were showing me designs for 12 cores. I mean, it’s really kind of great. I mean, I didn’t think we’d get here this fast. I didn’t think Moore’s Law would come into play. I mean, it’s the whole moving to multi-core for us is huge on our radar screen — and I’ll talk for a minute about virtualization and utilization as part of that — but I’m sure it is on yours as well.
We look at total cost of ownership, of course, right, and perspective from total cost of ownership for me a few years ago it was the server, and then suddenly things changed for us and we started to look at the entire rack. So, we moved from a perspective of, well, it’s just the server cost to, well, what’s the rack cost.
Now we’re looking at the total cost of a datacenter or a container, and actually that picture on the side of C-blocks is our sort of trademark for containers. Does everybody here know what containers are? Have you see containers? Are you reading about them? Yes? Anybody? Has anyone seen a container, walked through a container, investigated a container?
Containers are really, really cool, because they’re what it looks like in the picture. They’re sort of a big storage container that now is a datacenter.
So, these blocks are sort of self-contained units. They’re loaded up with racks of servers and network gear, and the right kind of cooling, whether it’s ambient air or water cooled, depending on where we place them, and we can wheel them in to facilities or you can wheel them into just the — and I’ll give you an example. Right now, I just came back from Boulder, Colorado, where we launched our first container. It’s part of an initiative with Virtual Earth. It’s outside in their parking garage, their parking lot area. And we’ve got a plug in for power and cooling and network egress. They’ve gotten about 100 times what they would have gotten in a datacenter storage environment, in a C-block, and it’s 100 percent wind powered.
So, you can do some very innovative things with containers as part of a total cost of ownership model for your company. Our Chicago datacenter, the whole first floor will be containers. We’ll move from about a standard of about 400, 450 watts per square foot to about 1,200 watts per square foot in a container.
So, when you look at the Chicago facility, for example, a standard datacenter, 30 megawatt datacenter, can fit about 25 to 30,000 servers. When you look at a container, a datacenter full of containers, we’re looking at about 300,000 — that’s amazing — 300,000 servers within the Chicago facility.
So, the numbers are off the charts in terms of the total cost of ownership you can deliver to your company when you think about moving your model from a server to a rack to a container to a datacenter model; really big numbers.
Power drop clearly is a huge consideration again because of the power costs, and then the whole price power performance balance, which we focus on a lot when we make these decisions.
Another key server trend again, and I know this is on your radar screens, is virtualization utilization. How many people actually think that you have a well utilized environment, getting more than 20 percent utilization?
Since you guys and ladies can’t look behind you, I saw one hand go up in the entire audience, one hand. I can’t see in the back.
So, I think that what we see in the industry is that we’re not really utilizing our environment as well as we could.
Last year, I went to a conference down in San Francisco that many of you may have heard of called Web 2.0, and I was really nervous about presenting at Web 2.0, because it was all — you know, here I come in, I’m Microsoft, it’s down in Silicon Valley, it’s all Open Source developers, and they’re developers of software, and I’m going to go in and talk about datacenters.
Now, you guys might find this topic interesting, but Web developers kind of go, oh, datacenters, not the hottest topic for me on a Tuesday morning, and I go in, and keynoting right before me are Eric Schmidt and Jeff Bezos. This is the O’Reilly conference, and O’Reilly says to both of them, you know, what’s your biggest challenge right now? And I thought they’d say cool products or market share or something like that, and they all said utilization. They said, you know, when I look at sort of building an effective company, and I look at my contribution margins and the way I need to build, I just feel like we’re not getting — you know, I think that Bezos said something like 15 percent utilization, which made me feel better, because we were like at 17 percent at the time. So, I was like, yeah, I’ve got them beat by 2 percent, that’s really good.
But we all have this issue, and that’s why I’m so thrilled to see Hyper-V and the stuff that you’ve been talking about over the past couple of days in the System Center solution, and we’ll talk about how we’re using that within my organization, but within server trends we absolutely put virtualization and utilization, using the stuff at optimal capacity, at the heart of things.
I’ll throw out a number, and don’t quote me on it because I’ll deny it, but I look at my budget between now and FY ’11, for example, and if I can move my utilization numbers up by — I’ll take a stab at 20 percent, I can give back to Microsoft upwards of $2 billion. That’s a pretty incredible number, right? I can give back to the company upwards of $2 billion. And I would imagine that within the scope of your world, by taking advantage of this technology, you can do the same; so, very big on our server trends, and things that we’re looking at.
In terms of network and trust, probably not a huge focus for this conference, but I do want to put it out there, right, because it’s so core. People, when they think about sort of TCO, they forget that the network is a huge part of this, and it’s a huge part of delivering, right, delivering scale and reach and also performance.
So, in my world I have to be able to deliver the same performance to millions of customers in downtown Manhattan, to four customers in the outskirts of Alaska, and their performance always has to be good and it has to be reliable. So, designing a flexible and agile network with scale and reach is really critical to performance.
We’re doing a lot in the networking space. I’m not going to spend time on it this morning, but I’ll share a little bit. You know, we’re doing a lot and expanding our broad global mesh, right, and making sure that we have fiber where we need it to deliver the kinds of performance to the 550 million MSN users and 350 million Hotmail users.
We’re also in the midst of deploying an edge network. So, we really are advancing our work in pushing out to the edge. We’re putting nodes around the globe at the edge, so that we can actually have better rein on our performance out to the customer, trying to get further out to that last mile. We’re working with some great innovations of peer-to-peer as well.
And then there’s the last piece, which is a really fascinating topic, which is datacenter virtualization, again not something at the network layer, not something we’re going to spend a lot of time on this morning, but very critical to our vision to delivering services across Microsoft.
So, the network piece is super important. Those technical evolutions we think are key and differentiating for delivery of services, so certainly a part of it, and then trust. If we can’t make sure it’s available, protect your Personally Identifiable Information, and make sure that you have adequate privacy, you shouldn’t use our products. So, trust and security is pinnacle to us.
Again, we’re using System Center to help realize some of that, right? Even something as simple as patch management, well, think about patch management to hundreds of thousands of servers or millions of servers. Think about the complexity of that. So, these tools become so critical as the environment gets larger. I can’t run my environment without tools like System Center; it’s not possible. It’s simply too large.
So, let me now pause for a minute, and turn my attention to manageability specifically. Again, because the environment is so large, so complex, I run — and I didn’t mention this earlier, but within GFS we run about 200 — over 200 internal services for Microsoft. So, we’ve got a couple of really large ones, and they’re the ones that you could already envision, and by that I mean Hotmail, Messenger, Spaces, search, and we call them our head properties, because they’re probably all in about 90 percent of the environment.
Then we have all these what we call tail properties, and they’re little things, because they only have like a couple thousand servers, right, or a couple hundred. I know, it seems silly, doesn’t it, a couple thousand servers? So, we have these tail properties.
So, you think about manageability from that perspective, hundreds of different properties, hundreds of thousands of servers. We really are committed to moving up the stack to Dynamic IT. It’s just it’s been a way of life for us for quite some time, and we’re really excited partnering with the System Center group to realize this vision together.
And we’re doing it from the ground up. And I’ll be honest, you know, and I said to Brad, I want to talk to the audience honestly — it’s sort of my MO at Microsoft, because I’m sort of WYSIWYG, what you see is what you get. I was in the back, they said, do you want makeup, I said, no, I don’t think these guys will notice. How do I look? Okay? Okay, good. (Applause.) Thank you. So, no makeup and no BS, right?
A couple of years ago I wanted to use our tools, but I really couldn’t, I really couldn’t. You know, we were built like very much — how many people here build their own tools, build your own stuff versus — or build some of this stuff or augment some of the stuff? Sure, you do. You know, because this is why Bob’s announcement and the message here at this conference is so important, right, that we’re finally — yeah, we’re delivering the right stuff in the datacenter. And I use it, you know?
We have a philosophy at Microsoft called dogfooding, you’ve probably heard about it, where we have to eat our own dog food, we have to use it. So, I’m really proud and happy to be able to say I’m loving the stuff coming out of the System Center group, and I can use it. And a couple of years ago I really — I mentioned to Brad, and before that Kirill, and I said, I can’t use it. Now I can.
So, we’re working together on a manageability vision where I dogfood it, and I’m testing this stuff in my datacenters, in the environment, and then sharing that feedback with Brad and the team, and we’re sort of iterating it into the products, and making sure that by the time you get these products, that they’re ready for primetime, and that they’ve been tested, not at small scale, not at medium scale, but at mass, mass scale.
So, the message here is, if they work for me, they’re going to work for you, because I’m testing them at volumes, and let me share some of the numbers that are kind of probably unheard of for many of you folks in the audience.
So, for us manageability is a philosophy if we’re going to be successful, and we’re doing optimal design, and we’re building absolutely from the ground up standards. You know, standards are — and people have glossed over the word standards, particularly at Microsoft, because they don’t have to run tens of thousands of diverse environments, but standards are so key.
But the reality is you don’t end up with a standard environment. Let’s face it, whether you have different people in your org building on different platforms or you go out and you buy a company, and your whole environment is Microsoft and then you go out and you buy a company, and they’re an Open Source company, and then, oh, now it’s not standard anymore, right, again that’s why Bob’s announcement on day one around an open System Center design where we can bring in different kinds of technologies, for me is so important.
You know we just went out and did a big acquisition of aQuantive, for example. We buy companies all the time. So, I get the environment where it’s kind of standard, and I’m running it, and then, boom, we buy a big company and they’ve got all kinds of stuff, and there’s Open Source stuff and Sun, and I’ve got to be able to integrate that.
So, from a manageability perspective, sure, design is great and this vision of standards is phenomenal, but it’s not realistic, not in the real world, not in our world where we have to actually run this stuff all the time.
So, I’m so excited about this announcement for you as our customers, as well as for me as our customer, because I am a customer of the System Center group, around an open model for manageability.
You know, back to efficiency, I’m put to test quite a bit in my environment, and we’re a very metrics driven company. So, I have to sit down in front of Bill and Steve and talk about my numbers. Without tools I can’t deliver the efficiency numbers that I’m held to as an executive at Microsoft.
You know, I’ll give you an example. We look at one measurement, which is number of servers per server engineer, and we have — again, we have all of these properties, and there’s one property — I don’t want to share too much — they’ve been very successful in automation. So, they have — they’re running — I’ll put it out there, because it’s too large an audience to make you guess, but right now they’re doing about 4 — no, they’re almost up to 5,000 servers per one server engineer. Whoa, right? Now, it’s a massive environment, this is a couple hundred thousand server environment that I’m talking about, but it’s 5,000 servers per one server engineer.
Then we have other groups that haven’t realized the vision of automation totally, and they’re at a couple hundred.
So, the cost, when you think about that from a critical for efficiency, for those of you — how many people here are actually held to a budget? Yeah, I’ll bet. No? No? Yeah, okay, good, right, I want your job, and then you can have my job.
But, sure, without the automation framework, there’s simply — who, here, catch — there’s simply no way that you could realize this stuff.
And so critical for efficiency. Again, you know, if you can’t — we have a philosophy in my organization, people always tease me and say I’m outwardly casual, inwardly rigorous. If you can’t measure it, you can’t manage it. And I’ll talk about some of the stuff we’re doing with operation center and gathering data so that we can manage and do capacity planning, but absolutely critical, and then all part of this automation vision, all part of moving up this curve with people, process, and technology toward Dynamic IT, and so — and in the datacenter, in the datacenter.
So, that’s where we’re going, that’s our vision, that’s what we’re partnering with Brad and the team on. And we’re getting there. I’m really excited because we are getting there.
You know, what do we get from System Center? I use the word “get” because within GFS we have a term, Growth, Efficiency, and Trust, and so that’s our mantra, Growth, Efficiency and Trust. If I can’t deliver that to my 200 clients within Microsoft, then I’ll be do something else somewhere, and I don’t want that, because I think the work we’re doing — I actually think I have the most fun and interesting job at Microsoft. I’m biased probably, but I do, I think that this is the most interesting space there is right now, very fascinating.
So, we get System Center. When you look at our growth numbers, when you look at the efficiency, and what we have to do in terms of PII and privacy, without System Center, without automation, without tooling, we can’t run it. So, it’s really — it’s a non-option.
So, let me talk about what we use within System Center, what we have used, and what we’re looking at.
So, within the System Center suite of — you know, and you could probably imagine this. Again, I have a whole team of people working on manageability, partnering with Brad, and the key things that we’re focused on are monitoring, of course, config management, and then utilization and virtualization. Again, this is of hundreds of thousands of servers.
So, let’s talk about monitoring today. Right now we use Operation Manager. I have — and you can probably envision this, you know, I have a big SOC, Service Operation Center. It’s usually during the day it’s about as dark as this room in there. People that you know come out and they’ll go, oh, is it raining or sunny out? But we do 24 hour a day, seven day a week monitoring. It’s the eyes and ears of our environment. And today we use Operations Manager as part of our SOC management tool, as the eyes and ears of our service desk.
I said that we were on a really, really large scale. Let me give you some perspective on the numbers here.
So, 80,000 performance counters collected per day. How many of you are hitting numbers like this? No? I’m getting giggles. No, I didn’t think so: 80,000 per day. A million events collected.
And here’s the number that really blew me away, because we kind of put them together as we move to the next version of Operation Manager. Look at this: A trillion rows — I didn’t make this number up, I can’t make this stuff up — a trillion rows of performance data per day. I mean, these are outrageous numbers. And, you know, we actually take this data and we use it, right? So, of course, we use the performance counters to make sure that we’re escalating problems and making sure that they’re resolved, but in terms of the events and the performance data, we take it back, we use it for capacity planning, we use it for a lot of predictive analysis within the environment. So, we’re not just monitoring, we’re actually using this data, as I’m sure you do when you use Operation Manager, to do all kinds of planning for the environment.
So, that’s where we are today. We’re really excited about Operation Manager 2007. So, what we’re doing, again in partnership with Brad, and we test this stuff, so we make sure that when you get it, it can scale to the hundreds of millions or trillions or billions, but numbers that you may never hit but we’re testing it for that, we’re piloting now System Center Operation Manager 2007.
So, here’s a number, and it kind of made me laugh, I said, I wonder if the audience will relate to this. We’re piloting on a thousand servers. I know, it’s crazy, right? That’s our pilot. That’s like, oh, let’s just do it on a couple servers, let’s like test it out on a thousand and see how it goes.
So, we started SP2 release, SP1 release candidate. We’re running this in a production environment, and then giving feedback. So, this is an iterative loop of feedback and partnership with the team on the pilot. And I have to tell you, we’re loving it. We love the solution. It’s working great for us.
So, now that we feel like we validate it in a small, 1,000 server environment, we’re going to up the scale a little bit, and we’re going to move forward with migrating 39,000 of the Operations Manager 2005 agents onto 2007. And if we get the performance that we’re seeing in the 1,000 server pilot, I think it’s going to blow the roof off our scalability and reliability numbers, at least what we have in our models for this.
So, we’re really, really excited about the pilot and about the solution set. We’re using it, we’re testing it at scale, and we’re really — you know, I’m sort of your testimonial –we’re really loving what we see.
Then, of course, there’s configuration management, and from my perspective at least config management is the heart and soul of everything. Again, when you have hundreds of thousands of servers in your environment, if you don’t have your config management right, it’s really impossible to run the environment.
You know, I’ll be honest again, we used to build our own tool, and I won’t tell you what it’s called, but we built — and we actually thought that we could do that, and it became impossible — you know, when we were a 10,000 server environment it was like, okay, we can run it. When we got to 50,000, 100,000, it became impossible to run, so we started to leverage and take advantage of Configuration Manager to data collect and report asset configuration or again our patch compliance, so we can maintain that security aspect of all those servers.
And now we’re looking forward, we’re actually starting a new endeavor now that we have access to Operation Manager 2007, we’re moving forward to bring this into the environment. So, we’ll be using it for asset management and other components of our configuration within GFS. So, again real time pilot, we’ll be testing it, pretty excited about the prospects, and so a huge partnership there.
The last piece I want to talk for a few minutes about is virtualization. So, we started to discuss virtualization and its impact on op-ex and scale, and as I said, honestly, you know, I think that we are highly focused on virtualization, but I don’t think we’re significantly further along than the Amazon concern or the GE concern around moving those numbers significantly.
So, within GFS we have those head properties, we have those tail properties, and when we look out to the tail, we have a program underway called Advantage. So, we said, you know, search, Hotmail, you’re so special right now you stay over there, and we’ll work on virtualization, utilization solutions for you separate from all the other hundred-and-plus properties that are all kind of different and all kind of nonstandard. We’re going to roll you all up into this Advantage program, and then we’re going to push the utilization there pretty hard.
So, we’re actually using Hyper-V now within the Advantage environment to start to move those numbers and try to move the utilization up significantly.
We set a goal of about — and don’t quote me on it again, but about 30 percent utilization in the first run of this. This is across the entire Advantage, very different product group, and I won’t get into what people have different definitions of virtualization and what is virtualization, so I mean all in how utilized, how well utilized is your entire environment.
So, we set a goal and we’re running this internal program, and it’s been a great partnership between System Center and GFS to actually use Hyper-V.
And I just want to emphasize, because I know that there are messages out there, oh, it’s been a long wait, I don’t know if it’s actually going to work, does it work, should I bring it in, shouldn’t I bring it in. I’m using it, I can tell you again testimonial, it works, and we’re really happy with what we’re seeing, and we’re going to expand it into other properties, including search, which is going to be our next endeavor to use Hyper-V in a very critical application that’s very mass and scale.
So, we’re pretty excited about it. We’re going to use, of course, Virtual Machine Manager to manage the environment, we’re already testing that. And the feedback we’re getting again from the product groups is phenomenal. We’re already seeing gains, and we’re really thrilled with what we’re seeing.
So, challenges: As I said, growth, efficiency and trust in the environment. We’re pushing hard with these products and solutions. We’re loving what we see. We expect to see more as we fully deploy these across the environment. I’m your testimonial. I think it’s been great. So, I strongly encourage you to take a strong look at Dynamic IT in the datacenter. I’m around for questions that you might have around how it’s working for us. But we’re excited and we hope you’re excited about the products and what they’re going to bring to your business.
So, thank you for your attention. I’m going to turn it back to our comedian who will bring us home. Thank you. (Applause.)