|
Remarks by Bill Gates
Intel eXCHANGE e-Business Conference
San Francisco, Calif.
Oct. 12, 2000
(Applause.)
BILL GATES: Well, good morning. I’m excited to be here and talk about the changes taking place in the datacenter. A lot of the great vision that we all have for the Internet assume that behind the scenes there will be incredibly reliable machines dealing with not millions but billions and billions of transactions, and doing it in a way that doesn’t require a lot of personnel and is done at very, very low cost. In other words, we need to bring the same type of miracle to the datacenter that we brought to the desktop. And that’s going to require incredible hardware advances and software advances.
This year, we’ve reached some amazing milestones in this that we call the Software Datacenter. So I want to talk about the progress we’ve made and some of the new things you’ll see in just the year ahead.
Of course, Microsoft and Intel have done a lot of things together. In a recent interview, Dr. Grove talked about us as "fellow travelers." What’s happened on the desktop is a phenomenon, and it’s a phenomenon that fed upon itself: as we got more volume, we got better tools, we got lower prices and we got more applications. And the more applications we got, the more incentive there was to drive the volume. So that virtuous cycle drove the desktop.
There’s a lot of exciting things still happening on the desktop. The PC is not the ultimate communications tool that it will become. There are areas of creativity that the PC is just now moving into. So there’s a lot still to be done on the desktop that will drive it and drive that architecture to evolve in connection with the things we do on the server.
We don’t want the servers not to be able to take advantage of the rich power of the offline capability of the PC, and so we need to meet all of these demanding requirements for the server while continuing to exploit that desktop device. And it’s that formula that can drive this industry growth to the same kind of percentage level that it’s been even in the early years.
Some of these evolving applications are very visible now: the idea of creating your photos, editing your photos and sharing those through the PC; being able to organize your music and share music. Of course, nothing new comes without a little bit of controversy in terms of exactly how that’s done.
It’s interesting. In the video area, the movie companies are moving very quickly to get their material available. They want to make sure that getting the material in an easy way, where the license fee is part of the habit of how you get it, is there from the very beginning. And so even though that data type is the most challenging for us in terms of the size -- the number of bits that need to flow -- there are some great breakthroughs there, not just about getting pre-recorded video, but also about taking your home videos, editing those and sharing those. In fact, that’s something we’ve got built into the latest consumer version of Windows, Windows Me.
We’re also very excited about the idea of new form factors for the PC. One, in particular, we call the "tablet PC" that you’ll take in with you to the meeting. As soon as you’re in the meeting it’ll notice who else is in the meeting, create a user group for follow-up, note taking. The number of hours in front of the PC can be increased quite substantially through this real-time collaboration and sharing and the new form factors that will allow you to really get rid of the divide you’ve had between your paper world and your digital world. By having the note taking, the information always available with the full screen device that’s connected up wirelessly, you can expand those scenarios and increase use and therefore value, quite dramatically.
For the first time, people are starting to think about the time spent in those meetings, to think about how people do analysis, how they do learning, and use this tool in a more profound way to improve every one of those things.
Now, as usual, we’re assuming miracle hardware advances. We’re assuming low-cost broadband first to business, but eventually into homes. We’re assuming wireless -- not only wide-area wireless, but wireless in the workplace and wireless in the home that connects things together. We’re assuming the PC form factor not only gets smaller and easier to move around, but also will have a high quality microphone and camera as part of that. And we’re assuming that it’s connected up to this Datacenter.
Now, all these activities are extremely visible, you know, the gyrations in the market, the level of investment, the excitement about these new things. The PC industry is not what it was in those early days where we had to go out and get people to pay attention. In fact, it’s kind of wild the way that the whole PC phenomenon is covered nowadays. I’ve got a little video clip I want to show that captures some of the ways that people are talking about PCs. So let’s take a look at that.
(McLaughlin Group and Judge Judy Video Presentation.)
(Laughter, applause.)
BILL GATES: Okay, so everybody’s talking about technology, and there are very high expectations for how all these things are going to work. Taking that back to the Datacenter, what do we have to do? Well, Data Center has been known for having lots and lots of capacity, centralized management and easy application deployment. The glass house is the glass house; everything operates there.
But there are some requirements that the datacenters in the past have not had that this new world required -- the ability to add scale at low cost, incremental scale, the ability to allow any system to fail and yet things stay up and online. The key difference here is that the datacenters of the past were largely batch oriented. In fact, in the few cases where they had to be real time, that is up all the time like a stock exchange or a phone system, they actually didn’t use mainframes or UNIX, they used very specialized systems from people like Tandem and Stratus. Those people had discovered the idea of software scale well before it came into the mainframe. But now with the new Windows capabilities, this idea of software scale is built in to every copy.
So you have to get multiple servers logically acting as one, the ability to manage it as one and yet have the performance and the fault recoverability that comes from the many different systems. We need lower costs. We need to take advantage of the desktop and we need to be able to write and update these applications very rapidly.
I want to emphasize that last point because the tools we’re talking about here are available to all companies, and so the competition will really be in who can take these tools and write applications around them, who can respond to the new business requirements in the most rapid fashion.
So the solution to this we talk about is the Software Datacenter. And all of those capabilities -- the dynamic scale, the manageability, the client exploitation, the great tools -- all of those are necessary elements.
Now, for us, this was a top priority in building Windows 2000. For us, the fact we ship Windows 2000 as a foundation this year is the start of changing the server world in the same way that the desktop world was changed. We want to do that, though, without giving up this PC utilization, the ability to replicate data without the user having to think about that, the ability to use the rich presentation, the low latency of working on that PC, to let people navigate that data.
I’ll talk a bit about XML today. It’s really the critical standard, changing the way that information flows to the PC, to not so much be presentation, which has come through HTML, and we’ll still support, but much more through XML. And so your ability to see data from multiple sites, to edit that data, to exercise your creativity around it and share it with other people, that’s the standard that’s key for the future.
The users want to be in control, in control of the very rich UI -- things like speech input, handwriting input, having the camera. None of those things work if you’re thinking about this as a terminal connected up to a large system. And you need the expandability that comes with having local execution and peripheral buses that make it easy to add in these new capabilities.
Now, the challenge to taking this architecture and moving it to the Datacenter has really been about scalability, reliability and manageability. And so I want to talk about each one of those things and what’s being done to address it.
With Windows 2000, we announced what was called Windows 2000 Datacenter. That shipped just last month, a very important milestone for us, because it’s the first time we’ve said we need a new level of partnership between the hardware vendors and Microsoft in configuring these systems. At the datacenter level, we need to limit the kind of drivers that are used. We need to create new support policies. We need to certify the configuration top to bottom, all the elements that are in there. We need to provide a single point of contact, including electronic monitoring of all of these systems, and we need to have lots and lots of professionals who are trained in exactly how to put these systems together and tell people if you want four nines here’s how you do it, if you want five nines here’s how you do it; all of the expertise that comes with putting those pieces together.
And so we have a number of partners that I’ve got listed here who joined into this datacenter program. And so it’s the first time in the industry where we’ve really emphasized the kind of variety, the availability of lots and lots of drivers. It’s the first time we’ve said, "Okay, if you narrow that down and focus on some top to bottom configuration, you can get the reliability that in the past you would have had to go to a single manufacturer to get." And so this I think is a very important step to balance things and get to these extremely high levels of reliability.
In terms of scale there are really two things that give us large scale. One is that the individual systems get more powerful. The wonderful advances in the microprocessor itself, the new memory bus work that’s coming with Infinibands, with the large packet sizes; all of those advances, including the Itanium processor family, the 64-bit advances, will make those individual systems far more powerful.
We’ve already seen the move from a single processor to a two processor to four and eight processor systems. Now we have partners like Compaq and Unisys who’ve moved up even to the 32-processor level in terms of their single system offerings.
So that dimension gives us very rapid improvements in scale. But we can multiply that by the ability to take those systems and combine them together into one logical system. And so those combinations deal with both the reliability issue and the scale issue, and we call that software scale and hardware scale working together.
So let’s take the most well known benchmark that’s out there. This is the TPCC benchmark, and it’s one of the best in terms of the way it’s audited by people coming in and looking at it. They force you to be clear about the cost and the availability and things like that. When we first tackled this back before Windows 2000, our performance was way below what the high-end UNIX systems could deliver. As we’ve gotten Windows 2000 software scale, the impact here has been very dramatic. That’s a period of less than a year. And by putting these systems together, this idea you can get this incredible scale, has really been proven out. And so today we are substantially ahead of any mainframe or any UNIX system.
These are the audited results. It was just this week that we announced the number one result here, which was the 500,000 transactions a minute and all of our results are in the -- with SQL Server, in that $20 per transaction range.
So if you look at the top ten here, before the announcement of Windows 2000 no PC architecture device was in the top ten in any way, shape or form. Now, we have six of the top ten. And it shows there’s quite a range here. You can go with a several million-dollar system and get 180,000 transactions a minute, or you can go all the way up to the $10 million system, the one that was done this week and get the 500,000 per minute. That gives headroom to any business out there. There is no Web site or business today that needs that level of transaction. But even if they do, it’s there, and that number we can keep driving it up simply by using the better hardware and adding more systems in. In fact, the way we’ve architected this thing, there really is no limit to how you can use that scale.
Now, this is not the only benchmark that we lead in. If you look at the SAP benchmarks, JD Edwards’ benchmarks, various benchmarks that are out there, PC architecture, which was always the leader in price performance, is now the leader in both price performance and absolute performance as well.
And particularly as those people use more and more partitioning you’ll see the gap on those things widen. Today it’s mostly Web sites that are using partitioning, and we’re working with all the LOB applications to get them to use that as well.
Now here are some examples of high volume Web sites that are actually using this architecture today. Even before Windows 2000, a lot of people had decided for their front end to go with this architecture, largely because of the tools and the low costs. And so in terms of penetration here’s a number of the surveys that are out there. The Windows system was way ahead of the others. We’re not content with this, because if you look at some of the high-end transaction sites, they weren’t using the architecture and it’s only this year’s milestones that will allow us to get all the new design wins to look at the advantages of what’s here.
The development tools are the most important part of this. It’s not just about saving money on the hardware system; it’s about being able to have the simpler application. Web sites, because tools weren’t optimized for letting people build Web sites, Web sites had been the most complicated pieces of software, hard to update, hard to keep reliable. The first generation of tools that are really focused on Web sites, thinks like our own Visual Studio.NET are what are coming out in this next year. And those things have the support for XML, the support for easy scaling and the ease of development that come with it.
XML has become the lingua franca that’s going to let us get these systems tied together. It’s not just a presentation standard that lets us do e-commerce. It’s not just a presentation standard that lets us do knowledge worker collaboration. And so for the first time, the release of the database and the release of the tools are built around XML.
That commitment to XML has a very deep impact on these systems, top to bottom, from the OS to the file system to the Microsoft Office. Every one of those things has been re-architected around the flexibility of XML. So although this year’s releases go one step in that, it’s just the beginning of what we can do there.
I thought it would be great to have one of these customers we’ve worked with on a high-end Web site come out and share a little bit of their experience. And one of the really exciting people we’ve worked with is GMAC. They’ve got a very complex business. And so I’d like to ask Niraj Patal to come out and tell us a little bit about his experience.
Welcome, Niraj.
(Applause.)
NIRAJ PATAL: First of all I wanted to thank Intel and Microsoft for giving me the opportunity to discuss our story. It’s such a compelling story that as you hear it, I’m sure you’re going to love it.
One of the things I wanted to start off with was we’re in the business of making very complex transactions. We’re not making very simple side transactions. The key here is, you know, we hear the word "mortgage" and I’m sure everybody here has gone through a residential mortgage and they’ve received a home loan, and I’m sure everybody can say that’s not the easiest process in the world today. But what we do is about a thousand times more complicated, because we’re only handling commercial mortgages. So what we call that is non-recourse lending, so we really have no recourse against the borrower. So the level of analysis that has to happen is incredible.
And at the beginning of the year the president of the company came to us, three individuals in our company and said, "Look, we have to revolutionize the way our business works today." Today the fastest loan that gets processed is about 90 to 120 days to get one commercial loan completed. He charged us with a goal to build a company and a platform in 90 days, which is shorter than what one commercial mortgage takes, which was an incredible goal to start off with.
So, you know, they put us in what we call the "skunk works." We were in one little building together. And as we were all trying to figure out how we’re going to accomplish this great task, the first thing that came to mind is, we need excellent strategic partners, not only on the technology side but the business side to achieve this type of goal.
And we went through all the basic players in the market, you know, Sun, Oracle, Microsoft, Intel, and we talked to every one of the partners that we wanted to bring to the table. And obviously we’ve selected Microsoft and Intel. That’s why I’m here.
But the story is very clear. We were able to achieve this business and this platform in that 90-day turnaround window. And we launched right away in March.
But to add a little pressure, our president actually added a little bit more pressure to our team, our vendors and us. We actually put in the Wall Street Journal about a month prior to the delivery of the product, 15 days straight there were full-page advertisements in the Wall Street Journal saying this is being launched. And that’s the type of pressure that was put on everyone, including our partners, because our partners’ logos were on those advertisements. So it wasn’t just GMAC at risk here, it was all of our partners and everyone.
And the story is incredible because here we are building a company, building a platform in less time than it took us to close a loan. And now we’re able to close loans in ten days, which is a very aggressive timeframe in the commercial world today.
One of the key differences with us was, we had to integrate back and forth between a lot of our partners, vendors and suppliers, and XML was a key strategy for us to go back and forth between all of our relationships.
We not only originate loans for ourselves, we originate loans for our partners, all the major investment banks in the market, life companies, you name it. So GMAC was always a part of being able to pass a loan where it needs to go. And we needed the flexibility and the scalability to pass this through to all of our lenders, let’s say, on our Web site. So that was a key part of our strategy.
One of the other things was, we wanted to make sure that the site was extremely reliable. How can I explain to a customer, who’s trying to get an $8 million loan, that my site will pick up? It’s just not allowed. How can you have, you know, trust in a company who’s trying to give you $8 million that, jeez, you can’t even run your Web site properly, which is kind of a statement to the platforms that we selected.
I’m happy to say that ever since the launch we have not had a single outage yet, which is incredible, because since March we’ve been running 100 percent.
Our next level expansion is kind of cool, because one of the things that we needed to do was expand out into our supply chain or information partners that we gather information from. And we’re actually aggressively moving out to the Pocket PC platform. We had a need that was very complex. We had tons of forms based on the property type, based on the investor, and what we needed to be able to do was the field inspector really doesn’t know what forms to use, really doesn’t understand what type of questions to answer when they’re out at the property. We needed to be able to pass this information directly to them on a handheld device, because using laptops doesn’t work when they have to climb the roof of a property. They needed a little handheld device that they can work with.
And selecting this platform is where we’re going, and our goal now is to actually condense our timeframe into five days for commercial loans. So you can see there’s an aggressive strategy here.
With that, I just can’t say anything more, but since the time the site’s been launched we’ve had about $2 billion worth of opportunities come through the site, just to show you that even a complex transaction can happen on the net and not just, you know, selling books can happen on the net.
So and that’s my story.
BILL GATES: Well, that’s fantastic. It’s been great working with you.
(Applause.)
BILL GATES: So I’ve talked about scalability and reliability, and let me now focus on manageability. A lot of people would say, "Okay, you have all these multiple systems. Aren’t your personnel costs proportional to the number of systems out there?" Well, there’s no reason for that to be true. It’s like saying, you know, isn’t your cost of using the PC proportional to the number of transistors in there. In fact, it requires great software design to make sure that you can see those systems in a very simple way.
The distributed environment that is necessary because of the ways these transactions will work in multiple partners requires a new level of software tools. You’ve got to have simple interfaces, interfaces that don’t overload you with information, the ability to reach out and see all the different things that are going on when there’s an event of importance.
So manageability, whether it’s thousands of desktop PCs or a Web site datacenter, is definitely an area we decided we needed to make a much higher priority. And there are several ways we dealt with that. First is in the Windows 2000 product itself, building in the instrumentation so all these different events -- "the disk is full," we’re getting an error on this communication’s link -- all those things can be bubbled up and then filtered according to the criteria that’s important for that application.
The infrastructure that’s in there, called WMI, is based on an industry standard for management. Active Directory is the ability to take policies, what user should be allowed to do what, what version of the application should systems have, and distribute that around and make sure that information is always available.
For people at the desktop, making sure that if they ever want to switch machines, if they lose files on their machines, making sure that that data can just come down automatically onto any new machine they plug in or log into, we also had to have that built in. So in effect, from a state point of view, it’s like having a terminal because the state logically is in the center, but you get the efficiency of having it on the PC. It’s a lot of basic things, including the terminal server remoting capability that just got built in to Windows itself.
As add-on products, systems management for the software distribution and then Application Center Server focus on this Web site where you have many, many different front ends and multiple backend systems.
But there was a piece missing, and that had to do with operations management. We went out, talked to our customers about what they were doing, talked to them about, you know, what their needs were and we found a great partner to dramatically accelerate what we needed to do here. And that was that we are announcing today, and this is a new thing that we’ve managed to keep confidential throughout the months and months of partnership discussion, we’re actually licensing from NetIQ their Operations Manager product. So this is a very important milestone for us in terms of having the capabilities that people expect out of a software datacenter.
So I’d like to ask up Ching-Fa Hwang and Jeffrey Snover that will talk to you a little bit about this new partnership and exactly what it means for the Software Datacenter.
(Applause.)
CHING-FA HWANG: We are very pleased to partner with Microsoft to provide the next generation operations management solution together. Let’s see a demo by Jeffrey to see our NetIQ’s operations manager in action. Jeffrey.
JEFFREY SNOVER: Thank you, Ching and thank you, Bill. I’m Jeffrey Snover, architect for the .NET management services and platforms. Today Bill talked to you about scalability and reliability. I’m here to talk to you about the last "-ity," manageability. I’m going to walk through three problems and show you how Microsoft is in a unique position to solve them.
The first problem is the fire hose problem. You go out and you buy yourself a server and we try and be helpful with that server by providing you all the information that’s going on with that server. Now, you take that and you multiple it in a scale-out environment times ten, times 100, times 1,000 and you’ve got the fire hose problem. Your administrators are getting tons of information fire-hosed at them and it makes it difficult for them to address the problem.
Now, you have a choice of how to deal with this. You could on the one hand just throw people at it. On the other hand, you could leverage Microsoft technologies to address the problem. Here we’re showing Operations Manager, licensed today from NetIQ, to show how to address that problem. Here we’re showing an environment that’s managing 100 or so machines. The product is actually capable of quite a bit more than that. NetIQ has customers that are managing thousands of servers in a single installation, processing tens of millions of events every single day.
To solve the fire hose problem, you have to centralize all the information about your resources. You’ve got to be able to see what’s going on with all the information at one particular point. But you also want to be able to see the information that matters to you. Many of you have hired, rolling out Windows 2000, have assigned an individual responsibility for the care and feeding of Active Directory. Here we’re showing that you can configure the powerful search and filtering capabilities of Operations Manager to take a look and show only the information about Active Directory for that individual.
When an error that requires an operator’s attention occurs, Operations Manager guides them through that, showing them the details of that information, allows them to assign it to an individual and track it through resolution. It also gives additional information about that problem, a re-description of the problem in English text. It tells you potential causes for that problem, potential resolutions to it and links to external data sources to get more information about that problem or about that area.
But perhaps what’s most impressive about the technology we’re showing you here today is what I have not shown you. And what I have not shown you is all the things that we automatically resolved without telling you about it. Operations Manager ships with business logic to be able to automatically detect common problems and resolve them automatically without requiring operator intervention.
Here we’re showing a list of in depth processing business logic to be able to manage Windows 2000 and the most common and popular applications, including SQL Server, Active Directory and Exchange.
So having shown you Operations Manager that coalesces all your information into a single place, we’ll go on and talk about the other things. But first let me say that NetIQ is not stopping here. We’re really excited about the plans NetIQ has in this area. They’re going to be extending this in three areas. First, when we ship our product, NetIQ is going to take that and extend it to other applications in your environment. So they’ll ship support for Oracle, SAP, Lotus. They’ll extend it to other environments. They’ll extend it to all the operating systems, Solaris, AIX, HPUX, Novell Netware, et cetera. And lastly, they’ll integrate it in with the common, most popular enterprise frameworks, so if you’ve invested in Tivoli, CA or HP, this solution will integrate into those environments without difficulty.
So having shown you the solution to the fire hose problem, that Operations Manager helps you efficiently manage your scale-out environment, now I’m going to talk about the next problem. That is the scavenger hunt. Your administrators have lots of tools. You have tools to do software distribution. You have tools to do inventory management. You have tools to take a look at your network. You have tools to do all sorts of things. But they don’t work together. And it causes inefficiency in your IT organization.
What we need is, we need to be able to take these tools, integrate them together into a single user interface, be able to have them talk together and then have them help one another.
What I’m showing you today is a technology preview of software developed at Microsoft that solves just that problem. Here we’re showing a Microsoft managing portal for a particular administrator. This tells him all the information he needs. At the top here it’s warning him, "Watch out; there’s a virus in the environment." Below that, there’s a status of the SMS rollout of Exchange 2000. This tells him all the resources that are affected by this change, when those resources are going to be affected and the team member to contact if they have any questions or concerns about this.
Below that is a report on the service level agreement performance of this team: are we meeting our objectives or are we not meeting our objectives?
On the next side we have that administrator’s calendar, email and task list.
All this information we merge together through the portal so that your administrators can be efficient and effective. All the information and tools that they need are right there at their fingertips.
The next problem to solve is what I call the playbook. If you have a new sports team, that sports team may be composed of lots of excellent individuals, but often they lose their first season. Now, why is that? And that’s because they have not learned how to work together as a team. Coaches understand this, and they’ve addressed this problem by publishing a playbook. A playbook says, "Here’s how we’re going to do things. When this circumstance arrives, here’s what we’re going to do. You’re doing to do this. You’re doing to do that. And here’s how we’re going to deal with that situation."
So too your IT organizations need a playbook, and now what I’m going to demonstrate for you is again a technology preview of provisioning software that has been developed at Microsoft to help you establish and then implement playbooks in your corporation.
Here we’re showing a provisioning of a user. Provisioning is when you get a new user, you’ve got make sure that they have all the resources and the environments configured that they can come and be effective in their organization. Often that is done by multiple people, often through different organizations and increasingly, through different companies required to provision a new user.
So in this demonstration, what I’m showing is that a new user comes in. They’ve already been assigned an account and been provisioned with standard desktop software. And now it’s come to me. There’s been a request for additional software for this user. And I’ll take a look at that, I’ll evaluate that, I’ll accept that request. It now moves on to the next step in the workflow and the next person in that chain is notified and the software keeps the people working together. So you see that with the playbook -- you have high quality, predictable IT processes.
So here I’ve showed you the resolution to the fire hose, the scavenger hunt and the playbook.
Just as in the past, Microsoft got serious about scalability and reliability and leveraged our unique vision and approach to surpass traditional approaches, so too Microsoft today is serious about manageability. We’re going to provide you the management technology and products for you to be successful.
Thank you.
(Applause.)
BILL GATES: We’re really pleased to have the partnership with NetIQ that’s really accelerating this critical issue in building these datacenters.
Well, there’s a lot of exciting things that this all leads to: thinking about software much more as a service than you have in the past, getting new knowledge worker tools that just automatically come down to your system, having a range of devices -- the full screen PC, the screen phone, the intelligent TV, all of those sharing the information so if you want to be notified about something it’s done and it’s only done when that’s appropriate. And you don’t have to manually move that information around; that’s done for you. The e-servers that everybody connects up to we’ll be using to provide not only more reliability, but more richness in this environment.
So XML is key under the cover. For the user, we’ve got to build new user interfaces that are far simpler than we have today, that take the rich graphics that take the multimedia and build those in.
And it’s because of these elements that we see as much visibility as the PC and the datacenter around that architecture get, we are just at the beginning of what we can do. And we’re very excited to be working on that with Intel and other companies in the industry.
Thank you.
(Applause.)
END
|
