MR. MARITZ:Good morning. I'd like to welcome all of you and thank you for joining us this morning in New York. I'd also like to say welcome and thank you to those who are joining us via satellite in Europe, and those of you who are joining us via satellites in other locations in the United States. This is one of a series of days that we've hosted to basically describe our strategies and bring you up-to-date on our progress from various aspects of the investment that we're making in computer software and computer architecture. As you may remember, we had a famous strategy day back on December 7th when we basically articulated our determination to build a major asset in the Internet space. And over the last year, we've done additional updates in terms of our Internet strategy, our Windows strategy, and today we're focusing on the very important part of the work that we're doing, which is to make sure that our underlying platform can scale up to handle the most demanding applications.

What we're trying to really accomplish today is to show you a great example of the tremendous power of the feedback loop in the investment cycle in the PC industry, which is now leveraging the fact that the microprocessing community can give us very, very powerful computing hardware to work with, and show you how we're building the software asset to match that.

We'd like to actually show you where we are in terms of the capabilities that are in the Windows-NT platform today, because this is not a new investment. This is an investment, in fact, that we've been working on for over eight years now. So we'd like to show you where we've reached, and show you customers actually using that platform, lay out a road map somewhere into the future, so you can see how we're going to continue those investments, and also discuss the necessary services and support that have to go with that complete strategy.

In terms of the agenda for today, Bill Gates will give us an overview of our overall strategy and approach and the philosophy behind these investments. I'll then come back up to give some more detail behind that, and the Gods of the weather permitting, we'll be joined by Dr. Hasso Plattner. I'm told that Hasso is now circling one of the New York airports trying to get permission to land. So, with some luck, he'll join us later in the morning. And then Deborah Willingham, who is our vice president of enterprise customer unit, will come up and discuss the very important investment that Microsoft and our partners are making in service and support.

I'd also like to take this occasion to recognize some very distinguished guests that we have here in the audience with us today because, as we said earlier, this is an example of the PC industry at work. It's not just a Microsoft phenomenon. So I would like to thank Robert Palmer, who is the chairman of Digital Equipment, who is with us here today. I'd like to thank also Roel Pieper, who is the CEO of Tandem, who is with us here today. And, additionally, I'd to thank Mike Perez , who is the vice president of Compaq's server division, who is with us; Wolfgang Dalichau, and Andy Orent, the vice presidents from NCR; and Sal Alini, who is the vice president -- senior vice president of the NT systems division at Amdahl.

So, with that, I'd like to ask Bill to come up and join us and take us through the first part of our agenda. Bill.

(Applause.)

MR. GATES:Thank you. PC performance has been improving at an incredible rate, both on an absolute basis and relative to other computer systems. And this is why today, at the desktop level, the PC model really is unchallenged. Even so-called UNIX workstations are in decline as PCs have taken over the top tasks, even real-time financial trading, high-end graphics, things that a few years ago you wouldn't have expected a PC to be able to do it's doing very well today. And that strong position just gets stronger in the next few years.

Now, this was predicted some time ago by the people who participate in the PC industry because the model is so much stronger. The level of investments and the scale just gives it a strength that no other part of the computing world has. As long ago as 1988, Andy Grove, who's not shy, predicted that his chips would become so powerful they'd be the heart of even the biggest computers. And it's really the great work that's being done by Digital and Intel that have given us engines that have allowed us not only to take the strong role on the desktop, but also to move up and be, by a substantial margin, the highest volume servers that are out there. If you look at the server world, Windows-NT is now playing a major role there. If you compare it to any individual form of UNIX, it's outshipping UNIX platforms by about a factor of five. If you take UNIX as a whole together, Windows-NT is outshipping it by more than a factor of two. And this momentum has built over the last few years as the tools and the applications and the hardware have come into place. And so Windows-NT Server is being used for very demanding applications, whether it's banking, manufacturing, running large distributed electronic mail systems, these are very demanding tasks that these over one million servers are out there taking care of.

Where are we on scalability? Well, it's today's theme. We made this a priority for Microsoft about eight months ago when we decided that the pieces really were in place to scale our systems better than any other. And so working with partners, we've been making some breakthroughs which you'll see here today, in scalability.

Now, one thing is already well established, for some key applications NT servers deliver better absolute work performance than anything out there. In electronic mail, file sharing and web sharing, we are the leader. In database, we are the price/performance leader. But in terms of absolute performance, we still have some work that we're doing to achieve that metric, and we're going to give you the road map that actually will get us to that level.

You're going to see two different things that come together in a powerful fashion. Symmetric multi-processing, so-called "SMP", capabilities that get faster and faster for a variety of reasons. The chips themselves get faster. The number of processors being used in these SMP configurations goes up, and the quality of software, whether it's NT or SQL server also improves a great deal. And so that's why you see the ramp being so incredibly steep. And then we add, on top of that, the clustering approach, and here's where you get even more dramatic results, is you can bring multiples of these high-power SMP clusters together and combine their performance to get very, very amazing systems.

Now, the key to the PC model is not just price/performance or absolute performance. It's the whole structure of how the companies working in the PC model work together. There's a lot of specialization. People who understand about making chips get to focus on that, and they get incredible volume there. People building applications, people offering support, the channels that we've built up over the years around the world are very, very important to this. But what it means for customers is that high-volume standard servers give them lots of choices. As they add new servers, they can look out there and say, who are they getting the best support from? Who are they getting the best price/performance from? Who has the best fault tolerance. And as they configure their system, they can mix and match using, in a branch office, smaller systems, using at headquarters larger systems, and those systems can come from different manufacturers because of the total compatibility that exists across the line here.

As you pick a PC platform, you know that not only do you get the best range of applications choices today, you know that the majority of software investments being made are being focused on that PC platform. And so you'll be able to take advantage of new things that go on in software. And that's the better pricing, better choices, better innovation going forward that applies across the board. If you look at peripherals and the price there, and the choices there. If you look at the number of certified professionals out there who understand the Windows-NT system versus any other system, it's much, much larger. Today, we have over 100,000 certified professionals who have passed Windows-NT exams, and over 10,000 solution providers bringing these things into the channel. So it's a very, very strong model.

To illustrate that, I wanted to show numerically what's behind Windows-NT, not just Microsoft and Intel, but a whole range of partners, many of whom are here with us today showing new milestones in terms of scalability. The application set is very, very broad, and the investments being made by those companies is pretty incredible. And so, it's dramatically larger than you'd find in any other system. Volume makes a big difference here. The ability to offer the software inexpensively, to get the incredible reliability into the system, and the breadth of applications all comes back to the volume. And that's always been at the center of why the PC has been so successful.

Here's a very direct comparison that's taking a very high-end server configuration. This is a configuration that Microsoft certainly buys a lot of, and it's four processors, a gig of memory, very large disk capability, over 30 gigabytes, and high-speed networking connections. And you see here it's over a $60,000 PC server, but it would be about two-and-a- half times as much configured using Sun Solaris. The very, very high margin here, of course, is in the pieces that you can't buy from somebody else. Such in the CPU, the Solaris CPU, and the Solaris add-ons CPUs are where you see the dramatic difference there. And it really goes back to volumes. The fixed R&D costs have to be thrown in somewhere, and that's, in this business model, where they get thrown in. In the case of NT, our software R&D investments are over four times as great as those being made in Solaris, but we can price NT substantially less because of the volume model, because of the scale that's actually out there.

Let's step into some of the key metrics about scaling. One is to say for a single computer how big is the database you can attach? Early in my career, a database that was 100k large was a very big database. Today when you buy a PC you get a gigabyte machine is just standard. And the biggest databases are 1,000 gigabytes or a terabyte. In the world today, you find a few dozen production databases that are a terabyte in size. So very, very few databases up at this level. If you take sales records of a company, a terabyte says that if you store 100 characters per transaction, you'd have to have over 10 billion transactions, which is more than almost any company has. So a terabyte is an incredible amount of information.

A few years back, Microsoft hired Jim Gray, and that was a great milestone for us because he'd worked on IBM System R. He'd gone on to product type that as their DB/2 product. Later went to Tandem to work on their transaction work. So, a few years ago, when he came to Microsoft and said he was going to help us achieve these highest levels, that was a great thing for us. And so his group has been working on building this terabyte server, putting together the single biggest NT node that's ever been built.

So, I'd like to ask Jim to come on out and show us what he's got here in his Microsoft Terra Store.

MR. GRAY: Hi, Bill. Good to see you.

MR. GATES:Good morning.

MR. GRAY: So what we've got here is the Terra Server. The Terra Server is a very large, one terabyte, database on a single NT node. This is one great big computer. The thing that's especially interesting about this is, this terabyte is entirely stored inside a SQL server. So, you know -- and what I'm going to do is walk you through the application. As I walk through the applications, we'll talk about various steps of the system.

So what I'd like to do is first, you know, you talked about how big a terabyte is. It's huge. (Laugher.) I mean, it just is amazing. If you go out into the Web and you take all the HTML pages in the Web and you drag them into your system, it's not a terabyte. If you go out to the New York Stock Exchange, and you take all the New York Stock Exchange transactions ever, all the stock exchange transactions ever, and store them in, I don't know, 200-byte blocks who sold, how much, half a terabyte.

So, it's hard to find a terabyte. It really is. And we wanted to find a terabyte that would be interesting to everybody everywhere, because we wanted to put this application up on the Web and make it available to everybody, because we want to show off our next generation SQL server and show off NT. So what we decided to do was to make a photograph of a planet, a photograph of the earth. And so, working with two partners, the USGS, U.S. Geological Survey, and also the Russian Space Agency, so the USGS is going to give us very good coverage of the United States, and the Russian Space Agency, incidentally, also has very good coverage of the United States. (Laughter.)

MR. GATES:I hope it's not better.

MR. GRAY: But they have coverage of a lot of other really interesting places as well. And so, we're getting the declassified space images from the Russian Space Agency, and also we've got the USGS things. And we've poured this into SQL server. And this is many terabytes. We're only taking the cream out of this data. This is many, many terabytes. And it's an interesting database because I think it's representative of the kind of terabyte databases we're going to see. It's hard to get a terabyte if you start storing character strings and numbers. You have to really get a lot of them, like the whole New York Stock Exchange is not a terabyte.

But the way disk prices are going and the way the Internet is exploding, a lot of people are going to have multimedia databases. And this is a multimedia database. This is not character and number records, this is picture records. So, we got this data from the Russian Space Agency, and their partner Aerial Images, and we put it in here. So now we've got a lot of data in the database. How are we going to find it?

So we took Encarta World Atlas, which is a really nice product, and we took its index, it's so-called Gazetteer, which knows about a million places on earth, and we took that Gazetteer and we put it into SQL, and so now we've got a million place names. It turns out that this is like a gigabyte of storage, a tiny little database. But you can look in this place names database and find pictures.

But we've got an in-line to the terraserver which is back in residence. So, Bill, where would you like to go today? Pick a place on earth, and we'll go visit.

MR. GATES:Okay. Let's take a look at Rome. I was there not too long ago.

MR. GRAY: Okay. So I'll type in the name Rome, and this is going to go off and run a SQL query, and come back with all of the Romes we've got, and there is probably Rome, Italy, that's what you're looking for.

MR. GATES:Sure is.

MR. GRAY: So that came to us over the Internet, actually -- it came to us over an ISDN line into the corporate intranet of Microsoft, and then whittled its way into the terraserver, came back, back came this picture. And this is very much of an overview. This is the big picture of Rome. And, again, this is the Russian Space Agency's eight-meter resolution imagery. So underneath this is a two-meter image. And we're going to zoom down to, that's the Vatican there, so we're getting closer. And we can zoom in again. This is really spiffy. We were running a little slower earlier. So we're zooming in on the Vatican. And that was the eight-meter resolution imagery, this is the two-meter resolution imagery. That's the plaza in front of the Vatican. And we could wander around, go down the street and look at it.

But each of these queries that we're getting, the idea that's gone on here is, we're sending the query off to a Web server, the Web server is sending it off to the SQL server, the SQL server is making an HTML page, one of these Internet pages, with this imagery on it, and sending it back to the Web server, and the Web server is sending it back to us over the Internet. We've taken all the imagery we've gotten broken up into little tiles that can be downloaded over the Internet in like 10 seconds on a modem. So far, so good?

MR. GATES:Absolutely. So, now, if somebody wants this image, I see a dollar sign there, can they buy it, or how does that work?

MR. GRAY: Exactly so. The Russians are not doing this because they're just nice guys. (Laughter.) They're putting this imagery on the Internet in hopes of selling it, selling the rights to use it. So, if we push this dollar sign, you can buy this image, and you can buy it in small, medium and large. Sort of like a Pizza Hut. Except this stuff is going to be delivered to you over the Internet. So this guy happens to know about my credit card, and your credit card. Our electronic wallet has been stored in this browser. So we're going to use mine.

MR. GATES:Okay.

MR. GRAY: Okay. So what's going on here is, we've set up a Microsoft site server which is able to take credit cards, go off and validate the credit cards, do the billing, and then tell the terraserver, hey, Jim's good for the money, give him the image. Okay? So I say, I want to purchase it, I like the deal. So, and incidentally, it's charging New York tax here. So that's the little step where I have to type in my password and really authenticate myself. And it's going to go off and decide if I'm good for the money, and say yeah. And so it's now ready to download the image. And so, where we go, we can download this thing. And that's going to cause that image to come to us over the Internet.

Now, it's going to take a while to do that, so let's go and look at the terraserver while that's downloading.

MR. GATES:So that's back at Microsoft headquarters?

MR. GRAY: Right.

MR. GATES:You can just zoom in on it?

MR. GRAY: Right. So let's search for the terraserver. Let's look for -- this is a sort of complicated SQL query. Let's look for lakes, and this is, again, using the Encarta World Atlas. We can look for airports, or cities, or mountains, or islands, but we're looking for lakes in Washington. And we're going to ask the SQL to find all those lakes, and then find the images that correspond to those lakes, and give us the pointers to them.

And we've got a bunch of images. One of the lakes is Lake Sammamish . So we're going to go to Sammamish, and this is not a Russian image, this is an American image. This is an aerial photo taken by the USGS, U.S. Geologic Survey. And we have all of Western Washington loaded into the database. And we're going to navigate over to your office first.

MR. GATES:So a little to the northwest of Lake Sammamish there.

MR. GRAY: Yeah, and hiking north a bit. And there it is, and we'll zoom in on it.

MR. GATES:Well, it looks like Microsoft.

MR. GRAY: Yes, and there's -- where is the campus, did I miss it? Maybe I need to go a little north. Yeah, there it is. Okay, so that's your office, Bill, I think, right in there somewhere.

MR. GATES:Yes.

MR. GRAY: Okay. And if we march north one tile, and there's cars in the parking lot, you can see that. You can't see the people. That's next year's images, I guess. And there's the building. This is an old picture.

MR. GATES:Yes. Building 18 is under construction there.

MR. GRAY: And so, if you went inside that building, what you would see is that. (Laughter.)

MR. GATES:The satellite can't do that yet.

MR. GRAY: The satellite can't do that. This is not a live link. So, that is the world's largest NT node. And what that is, is the box in the front -- these things are bigger than we are. I mean, that's one of the boxes, and there's like four of those. The box in the front is a Digital Alpha Server 4100. It's the little Digital alpha server. And it's got four processors, and two gigabytes of memory in it. And standing behind it are three great, big refrigerators full of disks. And those refrigerators have 324 disks in them, 1.3 terabytes of disk storage. These are Digital storage works disks. And then behind that, way in the background there, is a little tiny tape robot that stores two terabytes worth of tape. Okay? And we use that for backup and restoring. So that's the hardware picture.

And the software picture is that what we've been seeing, and it's been -- I'm actually impressed, it's been really fast because everybody in Redmond is sleeping, I guess, but the thing that's made this so fast is, we're sending requests to a Web server, the Web server is going on to a SQL server, and almost everything we've done -- in fact, everything we've done -- is a real complicated SQL query where it's going off and finding the records in the database, finding the images, pasting them together, and sending back a Web page through the Web server, and the Web server is the Microsoft Internet Information Server out of the box, NT. And back to this Web browser.

MR. GATES:Okay.

MR. GRAY: And when we decided to buy, there was another little Web server on the side, this guy here, and he's doing the financial stuff. And we're setting up a site server for the Russians and Aerial Images, their U.S. agent, and that's going to collect the money, and the money is going to flow back to the Russian space program. And we're also going to set up a site server for the USGS, the U.S. Geological Survey, and that money will flow back to the U.S. government. And then there was another server here that, if we wanted to use maps, we could navigate with that. Something we didn't show.

Database information, it really is a terabyte database. It's got a lot of records in it, 112 million records in round numbers. And if we go back and look, I think we've got the picture you bought, I bought, there's the picture of the Vatican. This is the Russian satellite image, and you can see here are some chairs in the Piazza, and here's a little accident on the bridge here. There's a bus. And so, what to say. This is the largest single NT node. It's a huge database. This is actually running the next version of SQL server. This is out of the box NT, but not yet out of the box SQL server. And it's definitely an information at your fingertips story.

MR. GATES:But bringing that kind of capacity, terabyte capacity, into industry standard computing, it just changes the economics of doing this kind of database altogether.

MR. GRAY: Right. When I go to a hospital and I see all those medical records stacked up and I say, gee, the cost of storing that and the cost of accessing it is just extraordinary. And if we could just put all of that stuff into a system, like this, it would both be a lot cheaper, and it would be a lot easier to get, a lot easier to access. It would just -- we chose this satellite imagery because there's a lot of it, and it's unencumbered, more or less.

MR. GATES:Great. That's fantastic. Thanks, Jim.

MR. GRAY: Thanks, Bill. Take care, bye.

(Applause.)

MR. GATES:So there we're showing a single node, scaling up to new heights. Now, what we want to talk about is how you can take nodes and get them working together. And this is where you get clustering coming in. Let's take a look, instead of just absolute database size, let's look at typical transaction rates. Catalogue order entry, where you might have 20,000 orders in a day, that would be a lot of transactions. New York Stock Exchange, where you might get into the millions of transactions, Visa, who can in a day can process tens-of-millions of transactions or the largest commercial transaction set, which is AT&T daily call lines, where they get up into the range of 200 million transactions. So these are the most demanding business applications. We decided to go quite a ways beyond that and to do a billion transactions in a day. So that's more than is done by any business today.

Now, we put this together because over the last several years we've been hiring a group of experts in transaction processing. And you've probably seen that we're now including the Microsoft Transaction Server as a standard part of Windows-NT. And this is a key foundation piece to let people do distributed applications. One of the members of this team, Robert Barnes, who was at DEC for over a decade and joined Microsoft over three years ago, took on the challenge of saying that once we have transaction servers, which is now shipping, that he could put together, using off the shelf hardware, something that went beyond any system that's ever been pulled together.

So I'd like to ask Robert to come out and show us what he's done with his billion transactions a day.

MR. BARNES: Hi. What we're going to show you is basically a global bank. We took the concept of starting with a region, say New York. And handling the daily transactions, people taking their ATM cards, walking up to an ATM, doing a deposit or a withdrawal. Now, it's nice -- as Jim showed, we can build a single system to handle a lot of data, but what happens if the amount of transactions that you have to handle is bigger than any one system could handle? Well, you need a way to be able to bring in multiple systems, tie them together with transactions, because transactions make the work that spans the systems reliable. You can count on failures not breaking the system.

So we're going to see on the display here, we've got a world map. And as, say, the New York area starts to serve transactions we're going to see that part of the map highlight. We've got some shelves over here and on those shelves we're going to see servers appear, as those servers start handling transactions. The bank that we've built actually has 1.6 billion accounts. That's probably enough for every family in the world, or about a quarter of the population.

As the transactions start happening we're going to see a transaction rate here, measured in transactions per second. We're also going to see a time-based graph that will show our progress as we stage these transactions. In other words, as we add a server, we're going to see more performance. That's the idea of showing more scaling. What's really important, at least to me, is this red line. Our goal was to do some staggering amount of work, to really capture the imagination of people saying, this is big. That line represents the rate which, if maintained for 24 hours, you'll do a billion transactions in a day. And what's important to me is that we've actually done this, several times.

The equipment that we use to do this is actually here in the building. It was a little too large to fit on the stage. So we converted an art gallery upstairs into a data center. And we used, as you mentioned, off the shelf hardware, all of the software we used to build this is software that's shipping today, Windows-NT, SQL Server and Microsoft Transaction Server.

So the last thing I want to point out is that we have a counter here for the number of transactions. So we've been talking about ATM transactions, let's actually do one. Okay. Come on over here to this ATM. Now, this ATM looks like a regular ATM, but what's interesting about this NCR ATM is that it's running NT Workstation. Go ahead and enter the PIN, collect some cash. You're going to get a receipt back. Now, what's happened since the time you requested the cash is that a request went up to the server, here in the lobby, we updated the balances for the teller, for the branch and for your account, authorized the transaction and returned the authorization to dispense cash. You're going to get a receipt.

MR. GATES:I got my cash.

MR. BARNES: You got your cash. You can keep that. And NCR money, on your receipt will actually be the response time, measured in milliseconds that it took to do all that work. So only 20 milliseconds when the system is just dedicated for me. That's good though, that's still under a second. That's fantastic.

Now, I'd like to point out that now the counter shows we've done one transaction. You and I are going to be able to do a billion. We need a synthetic load here. I think we'd better get some help. I'm going to start the load. What we're going to see is a server appears as transactions are being generated. We'll notice that the New York area has been highlighted and we should start to see some transactions here shortly.

While we're waiting for this to get going, we've actually got a tape of Jim Gray doing a tour of the data center. So why don't we have Jim do that and we'll hear more about the system that's running.

(Video Break.)

MR. BARNES: Great. We had a little bit of network difficulty, so we had to actually restart. As you can see, we've done much more than one transaction. Our simulated load is starting to happen. But, we see that the North American Region is actually being served with about eight servers. If we look down here at the transaction rate, we're doing about 7,000 transactions per second. Measured in terms of billions of transactions per day, we're at about .6 billion transactions per day.

MR. GATES:Now, new servers are coming on line, the synthetic load is just ramping up and ramping up, until you get all 20 servers running transactions.

MR. BARNES: Right. Once all 20 servers are up and running we're going to see the random work load that we've got generated. Now, you notice the line isn't going to be necessarily totally straight, because this is real work. In a real environment you don't get flat lines, you get random amounts of work. Now, while this incredible rate is going -- in fact, why don't we wait until we get to the rate for a billion transactions per day. We'll go over and do another transaction, see how our response time is then.

What's important for a transaction processing system is to be able to handle transactions under a load. And what you'd expect is that the load should be predictable. So hopefully we'll be getting a response time here of under a second. I still have enough money, I guess, I hope so. Yes, it took about a half-second. That's great. And you'll notice now, we're actually at the rate to do a billion transactions per day. We're at -- between 10,000 and 13,000 transactions per second. When this gets to steady state, in other words when all the network connections have been made, and all of the work is being generated, we average somewhere between 10,000 and 14,000 transactions per second. We've run this for over 24 hours and done over a billion transactions.

Now, another important aspect of doing transaction processing is what it costs to do it. Let's do some quick calculations here. If we assume that the hardware and software to do this benchmark, what's running up in the data center is about $5 million. And then we took the rate of transactions we're doing, which is a billion a day, we ran that out three years, that's a trillion transactions. You take the $5 million system cost, and a trillion transactions, that ends up being 5 microdollars, or 5 millionths of a dollar per transaction.

And the way this is designed, you can scale it up even further. There's nothing magical about the billion transactions. If you want a bigger number, you decide, you know, the vendor you're going to order your equipment for, you place your order, you bring the trucks of drives in, you plug them together, and you can run.

MR. GATES:Well, that's fantastic. I hope people will take a chance to actually see it in action. It's a pretty impressive system to see all those disk drives chugging along, doing the different transactions. You really sort of have to feel it to experience the full power.

MR. BARNES: So it's a whole new definition for a PC. It's not your average PC, but, they are PCs and they're running the same software that people can get today and it's equipment that people can order today. What I think is important about this is, for us this was impossible three years ago. It's possible now, as we're showing. And we're going to make it even easier for people to build these kinds of apps in the future.

What I'd really like to finish with is, for those people who say NT doesn't scale, I'd like to invite them to visit my data center.

MR. GATES:Okay. Thanks, Robert.

So it's different levels of transaction performance. There's different systems you combine that fit into the Windows family. One thing you can say for sure is that for one-way, two-way, four-way and now with eight-way type systems, NT equipment provides the best performance and the best price performance. Once you get up above eight-way, that's where clustering needs to kick in. And as we're bringing that to the market later this year, you're going to see record numbers up in those areas as well. But, the eight-way systems can deal with well over 95 percent of the market that's out there. In fact, if you look at Sun's volumes, less than five percent of the systems they ship are up in that area, where they still have reasonable transaction comparisons. And so we took systems at each level there, and you can see in every case either twice the performance or half the price for equivalent number of transactions.

Now, the product we're delivering this with is our BackOffice product. It includes SQL Server and many of the other pieces that go into building these high end systems. We partition BackOffice in the same way we do with NT, where you have a small business edition, that builds everything in and is very easy to configure, a standard edition, which is for departmental servers, and then the enterprise edition, that has the highest performance elements and the maximum scaling capability.

Our view is that although in the past people used to think of applications as being in two buckets, operational applications and then productivity applications. But, the boundary between those is going to go away. If you're a company that really likes to track your sales, know what's going on with customers, you want your operational systems and your office productivity systems to be tied together. And so your web site is publishing, has collaboration with customers, but it's also order taking and it's something you want to analyze. You want to know, who's coming into my web site, which of the people who came in ordered. If I was trying out a special offer with some of those customers, how much of a difference did that make?

And so there will be this rich, rich mine of information coming off of that web server and being used by everybody inside the company. And so the range of these applications can do things that applications in the past simply wouldn't have tackled. It wouldn't have been possible. The hardware would have been too expensive. And that's why you've had boundaries between these applications in the past, which held people back.

So Microsoft is investing in four key areas. First, and what we're highlighting today, is scalability. We're moving very, very rapidly. People are going -- people are very surprised at how quickly we have gotten to these levels of performance. But, we're not stopping here. We're going on and on, so that no one will have to worry about the headroom of their PC-based server systems.

Another key focus is interoperability. We understand that even though a very high percentage of new systems are using our technology, customers want to retain the applications and hardware they have and have great interoperability. We've always promoted initiatives like ODBC or OLEDB, that connect databases together. We have an SNA Server product that's focused on connecting to the mainframe. And then back in our pavilion we'll be showing the SNA Server doing some new things that were impossible before. We have a connection, the CICS, where through our transaction server you can easily connect up to CICS applications. And we'll also be showing VSAM connectivity, which is a new thing. So a lot of milestones in the area of interoperability.

Another key area is manageability. We're not focused on this today, in terms of talking about what we're doing. In fact, we have an event that will be back at our headquarters about a month from now that will take a particular focus on manageability. This is an important area. It's really the only thing I'd say that's more important for us than scalability.

And under the term zero admin Windows, we have made some breakthroughs, things that will allow you to have the benefit of a portable machine that you can go off and work with flexibly, while still having all the things you do be manageable and saved up on the server. And so you can go to different machines, you can replace a machine and you don't ever have to worry about the variety of state that nowadays is out there on the PC level. So the advance in the directory, the advances in the way we do storage management and zero admin Windows are really going to change the manageability story for PCs.

Finally, and also of key importance, is what we're doing with partners to get the services for capacity out there. Whether it's transferring knowledge to partners, putting together consulting groups, blueprints, growing our own Microsoft consulting capability and really working through what we call solution partners, that's been an incredibly successful strategy for us. And a lot of those solution partners are here today, showing the works that they're doing.

One way we thought that this would all become very concrete for you is to talk about it in terms of the opportunity. Now, we think of the Internet and how the world would be very electronic, transactions will take place that way. Well, in order for that to come together, we have to have scaling even beyond what we're showing today. And here's the road map, going from symmetric multiprocessing to clusters, and then moving up to distributed clusters, and not only distributed clusters, but distributed clusters that can take advantage of the performance of the PC. And so you can move applications around very easily and use any idle computing cycle. Having the same operating system on the desktop machine and the server machine enables that. But, in order to find those spare cycles and easily go out and use them, those are some advances that we are working on today, and fit perfectly into our Windows distributed framework.

This is all very necessary to get the Internet to achieve its full potential. And so there is a lot of opportunity here. Using the PC investment cycle to address all applications, allowing businesses of all sizes to not worry about the technology they buy running out of gas, letting people do new things, where you can see information in a richer way than you ever would have before. So this all fits into the incredible Internet explosion and the need for businesses to work with more information.

In fact, what I'd like to do now is take some of the comments our customers are making about NT and scalability and let you hear those. So we put together a video of some very exciting customer examples and let's go ahead and roll that. Thank you.

(End of tape.)

Question & Answer Session with Bill Gates, Paul Maritz and Deborah Willingham
Read Paul Maritz's Transcript
Read Deborah Willingham's Transcript