Launch of Microsoft SQL Server 7.0
Microsoft president Steve Ballmer
November 16, 1998
[Due to the varying sound quality and subject matter of tapes, the information in this transcript may contain inaccuracies.]
Mr. Ballmer: Well, let me start by welcoming everybody here today in our live audience, shall I say, in Las Vegas.The people in Las Vegas, I know, have been struggling with buses and traffic to get over from the convention center.So, I say an extra special thank you to you, and to all the thousands of people who are joining us over the satellite in over 35 cities around the United States, to you also I say thanks very much for taking the time to be with us today.
This is an interesting speech for me, I’ve got to admit.This is the first speech I think I, in my life, I can fairly say I’ve given twice in one day.We actually did the SQL Server 7 launch event this morning in here about 7:00 in the morning.It was the first time we’ve launched a product in Europe, which is where we were launching it this morning, before we launched it in the United States.But literally every seat in this ballroom was absolutely empty.So, I’m particularly thankful that you’re giving me your real live human time during the U.S. launch of SQL Server 7.
This is an exciting day for us, and I hope and think an important day for our customers.SQL Server 7 is really a product that was designed from the get-go inside Microsoft as a breakthrough.A breakthrough in the sense that it would allow PC servers to scale up and take a position running applications in enterprises large and small that it never had before.And a breakthrough in terms of delivering new capabilities and new approaches, and new facilities that have not been available in databases on any operating system or any piece of hardware at any time.
This is a product that I’m afraid we probably won’t do justice to today.It’s so rich.It has so much in the way of new capability, new tools, new ability to take care of itself so that you can spend less time and less effort having to manage and administer databases that I’m just not sure we’re going to do it justice.And I think it’s part of creating a real track record and history and presence for Microsoft in the database business, the pioneering work we’ve tried to do to bring the PC server into the database mold and Windows NT; the first introduction that we’ve done of graphical user interface tools, to be used managing SQL databases; the work that you’ll hear about today where we take the first steps to really integrate data transformation services, and OLAP services directly into the database; the work that we’re trying to do to open up the world of data to many more people through the integration and front-ending of the data through Office 2000, and that’s just a small part of the list.Scalability, record locking, availability, blah, blah, blah, blah.So, I’m not sure we’re going to do full justice to the product today in the short time that we have, but I think it’s important for you if you leave with one thought, to really understand that this is a breakthrough product that we encourage you to take the time to get to know more about and learn more about.
I think our approach in building this product really started with the right two ingredients.The first was making sure that we really had a team of people at Microsoft to lead this effort, and lead it in such a way that we could learn from all of the experiences of the past on mainframe systems, on Tandem
systems, on minicomputer systems, and to get the best minds from industry, from academia, to really help in the creation of a platform to run the most important applications in the business.
I actually think our investment in this team started 10 years ago, when we brought Dave Cutler from Digital to Microsoft to start up the Windows NT effort.Jim Allchin, who is a pioneer in distributed systems joined us in 1991.David Vaskevitch, who’s been really a leader in database technology and architecture, application architecture, joined in 1993, and since then a host of very talented people, Peter Spiro, Hal Berenson, Phil Bernstein, Paul Flessner, Casey Kiernan, and many others, with talent and experience relevant in a variety of different spots in the database industry.In a way, perhaps, what I would say the Godfather to the entire effort was Jim Gray, who joined our research staff, who’s really been involved in most of the pioneering efforts in the database field over the last 30 years.
So, we think we started the right way.We went out and benefited from and brought in the most talented people in the database industry to really lead this effort.We’ve had some good success in the market.We’ve made some good progress with the technologies that proceeded under the leadership of Ron Sukup and some of our internally talented people that had grown up.But it was time to set a bold new step forward, a place where we could really deliver a product that I think would find its true place as one of the leaders in the database world.And I think we have absolutely the right team of people to lead that effort.
The second thing, the second key element of the starting point for SQL Server 7.0 was to really take the right approach in terms of listening and focusing in on the core themes, and the core areas of greatest interest and importance to our customers.And in a sense, I guess, we break that into three things.Number one, and this builds from the heritage that we had as a company, we needed to take steps to make the database world easier — easier for end users, easier for developers and easier for database administrators.
We had to, though, go outside of ourselves and listen and respond to customers’ insistence on scalability, not only a scalability up, because there’s certainly been a lot of focus in the industry on the scalability up of SQL Server and the PC server in general, but also scalability down, so that we have a system that was administerable, manageable and reasonable for smaller businesses and for the laptop of the mobile users in many large companies.
And, third, we wanted to make sure that we focused in on what I think I hear from IT people and business people as one of the core issues of our time, that’s this notion of business intelligence.There’s more users in more companies around the world who are frustrated to know that their companies have the information that could help them make better decisions, yet are unable to open up that world of information to the users in an important way.
So, these were kind of the three customer inputs that we took as most important.And that drove us to do a long list of things.This is just a small subset: row level locking to give us additional scalability and familiarity for people who have written applications for other databases, update replication, Windows 95 and 98 support for people who wanted to build mobile applications, heterogeneous query because we know we’re living in a world where not only is Oracle data and DB2 data important, but so are data that are stored in non-relational systems also important.We built in the data transformation technology, and integrated in the OLAP technology that can make this a rich business intelligence solution.And whether it’s for scalability, whether it’s for business intelligence, or whether it’s just to make these systems better for end users, we knew we needed to deliver increased performance.
So, the right people taking, I think, very much the right focused input from our customer base.Now, this turns out to dovetail very nicely with a dialogue that we’ve entered into with a large number of our customers around the globe about where the value of information technology will come.This is a big issue — 50 percent of all capital spending in the United States today goes into information technology.Yet, most business people still don’t have a very good gut judgment about where the value comes out of information technology.
About a year-and-a-half ago, to help really give the business people some indication, some aspiration on what the value of IT could be, Bill Gates started talking about a concept that we call the digital nervous system.And if you think about a company the way you might think about the human body, every company has a nervous system.It helps me think, it helps me take input, it helps me communicate, analyze, plan, and take action.
And the question is, how do we let this incredible investment in information technology benefit the business?So, we focused in three areas, business operations, the business of the business; managing knowledge and putting it in the hands and in the fingertips of the people who need to take action; and electronic commerce, opening up businesses large and small to their customers for improved selling and customer service.In every one of these scenarios, not only is our Windows product important, and our Office product important, but so, too, is our SQL Server
product important.And if you listen and hear about the kinds of operational things, commerce things, and intelligence of knowledge management things that customers want to do, the things I talked about from a database perspective, and the things that I think we see from our business customers from a business perspective align.
So, what I want to do now is take a few minutes and sketch through how the many improvements in SQL Server Version 7 help address some of the kinds of applications people want to deploy in these three areas.
The first one I want to highlight is business operations.And when we say business operations, we mean running the business of the business — the accounting, the financials, the manufacturing, the ERP systems, the customer service systems.This is the traditional, at least in larger companies, purview of the minicomputer and the mainframe.But with our absolute conviction in the kind of quality of work that Intel and Compaq and Hewlett-Packard and others are doing, we know that PC systems will be the most powerful, the most scalable systems in the world.And we need to provide the operating systems, and the database that really allows that power to flourish.
So, we’ve focused in our attention in doing so in the business operations space.With SQL Server 7, we’ve made major progress in a number of areas.First, and I’m going to show you some numbers on this, SQL Server 7 is far more scalable than our Version 6.5 product.Scalability is only important, though, if you give people the tools to keep these systems up seven days a week, 24 hours a day.And we’ve made a lot of improvements in availability so that, for example, you can provide
you can perform online maintenance on the database while you still have people using it for transactional or decision support purposes, and I’m going to come back and highlight a number of the improvements we made in that area.
From an applications perspective, again, a huge leap forward.There will be over 3,000 applications running on top of SQL Server Version 7 within the next 12 months or so.We’ve also done some things, though, we think are very unique and appropriate to this area.The first is in the area of auto- configuration and auto-tuning.I’m going to explain those in more detail, but for now suffice it to say, we think it’s very important to be able to put a SQL Server someplace and have it largely maintain itself, take care of itself — whether we’re talking about the remote sales office, or distribution center, or plant, or branch of a large company, or whether we’re talking about a smaller organization — the cost, the complexity of bringing in database administrators cannot be prohibitive for the adoption of these technologies.
And we think we’ve really broken through in a unique way in terms of this approach.We also think that that leads to a lower cost of ownership in some very important ways.
What I’d like to do now is show you a little bit of a video from one of our customers, Elcor Corporation, who already today is live with SQL Server 7, a beta release, but they are live running the PeopleSoft human resource software inside Elcor Corporation.Thank you very much to the folks from PeopleSoft, who are live running SQL Server 7 and People Soft at Elcor.
So let’s roll the videotape, please.
Mr. Ballmer: I think that gives you a pretty good sense of the kind of work that at least one company is already benefiting from with some of these popular new line-of-business applications, operational applications on the SQL Server 7 platform.
I want to come back, as I promised, to this notion of auto- everything.One of the great problems and issues today is the expense and complexity of taking care of databases, and I don’t care how large or how small the company is.I was in Manhattan recently, and I was talking to some CIOs of some of the Wall Street firms, and they say today because of the scarcity of supply, and the great need for database administration talent, they might have to pay up to $250,000 a year for a database administrator for their UNIX database systems.Absolutely anything, whether it’s in that size firm down to much smaller companies, that we can do to make the database take care of itself is important.
SQL Server 7 supports on-demand memory, SQL Server 7 will automatically grow and shrink database size.You can set a minimum, you can set a maximum, SQL Server 7 will do the rest.SQL Server 7 automatically goes out and collects all of the important operating statistics about your application, and configures itself appropriately for the behavior that it sees in your application.It tunes itself.It will watch the specific queries and transactions that your applications make, and it will go back and tune itself for better performance.We have a special index-tuning wizard specifically to improve a lot of the query performance with the new query optimizer that is built into SQL Server 7.
For the smallest companies this is essential; for the mobile user, this is essential; for the largest company, this is just awfully damn important.And frankly, we think SQL Server has really
— with SQL Server 7 we’re really in a place that I think everybody else in the database industry will want to be over the course of the next several years.
Scalability has been an issue that people talk about a lot in the database industry.In particular, people talk a lot about it in context of Microsoft, and people growing up from the PC world.SQL Server 6.5, our initial release, our last release, pretty good product.But, if you take a look at the improvements that we’ve made in scalability between 6.5 and 7.0, it’s amazing.We went to three of the biggest
— the three biggest ERP vendors, SAP, Baan and PeopleSoft — and we said, look, this product isn’t going to be shipped until we’re scalable running your applications.So we can all feel comfortable going forward to the market and talking about this.
SQL Server 7 has over doubled the performance of SAP R3R/3, running on NT and SQL Server with the tuning and work that’s gone in.In the case of the Baan application, as measured with their standard benchmark, Baan reference users, we’re over 10 times, 10 times more scalable with SQL Server 7 than we were with 6.5.We’re almost four times more scalable running the PeopleSoft human resources standard benchmarks.This is breakthrough.This is important.It’s not only important for SQL Server; I think it’s important for really challenging this whole notion that says, the PC server foundation can’t be rich enough to run entire enterprises.
Well, you might look at these numbers and say, it looks like great improvement, how bad were you before?Microsoft must have been pathetic.Well, I think the key thing to focus in on it how good are we today.We still
Mr. Ballmer: I’m a salesman.We still
— and I will admit this, won’t scale up to attack the very largest problems in the very largest companies.But, SAP, Baan and PeopleSoft all validate that with these numbers, over 95 percent of every installation on all of these systems could be run entirely on NT and SQL Server, over 95 percent.So the largest of the large, not all of them, but damn close, can run entirely on PC servers, with SQL Server.Baan, for example, is so excited about this, that together we’re working on really driving SQL Server to be 90 or 95 percent of their overall volume over the course of the next 12 to 18 months.And so we’re excited about the kind of partnerships.
I think the ERP applications are particularly important, because for many companies they’re the ultimate test and benchmark of what it means to be an enterprise application.And I think this kind of performance certainly says that SQL Server has the scalability — we still have to prove reliability and availability — but certainly has the scalability to tackle the largest enterprise needs.This is really a critical issue for us, and we’re very excited about where we are.
Rather than just have me tell you about the product, we thought we’d have some customers join us and tell you about their experiences deploying SQL Server 7 in its beta form, into production.So I’m going to invite now to join me on stage Britt Mayo.Britt is the director of IT for Pennzoil Corporation, a $2.7 billion energy company, and well known for the Jiffy Lube.
Mr. Mayo: :Hey, Steve.
Mr. Ballmer: Hey, Britt.
Mr. Mayo: :Great to see you, as always, but especially when we get the opportunity to talk about another case where Microsoft and Pennzoil have worked together to create a great result for our companies.I hope you don’t mind, but I want to work off of this podium, because I never could resist a chance for a free product plug.
Mr. Ballmer: Go ahead, Britt, it’s all yours.
Mr. Mayo: :Thanks, Steve.
Pennzoil is 109 years old now.And a little background about the company, we explore for oil and gas worldwide.Our Jiffy Lube subsidiary that Steve mentioned is the leading franchise for an operator of fast oil change centers.We’ve got 1,600 retail outlets and are growing.And we manufacture and market a variety of lubricant and automotive products.The most well known being our bright yellow bottle, which has been the number one motor oil for the last 13 years.
Now, as you know, Steve, getting to number one takes a lot of hard work.But, to stay there you have to work even harder, and always be alert for ways to change your business, and do you business better.And so change has been a constant at Pennzoil, and that’s never been more true than this year.
Over the next few months we’re planning to merge our manufacturing, marketing and Jiffy Lube operations with Quaker State corporation, and create Pennzoil-Quaker State company, the world’s premier supplier of consumer automotive products and services.Then we’re going to take our upstream oil and gas business, and turn that into Pennzenergy, which day one will be one of the 10 largest EMP companies in the United States.
So basically in that environment, in ’93 we set out to create a world- class computing infrastructure by moving our production processing from our IBM mainframe data center to a Microsoft-based infrastructure.And as we did that there were a lot of projects we did over the past five years, but probably none bigger or more significant than implementing virtually every module of SAP’s R/3 system.We went live on SQL Server 6.5 in November of ’96.
Mr. Ballmer: I remember.
Mr. Mayo: :Yes, and we scaled up to our current size in February of this year, you probably remember that one, too.
Mr. Ballmer: I remember that, too.
Mr. Mayo: :So, basically, as the merger proceeds, what we need to do is we need to keep the operations running smoothly, we need to maintain timely access to the operational and sales data, to and from thousands of locations.And we need to do that while keeping the cost savings that we’ve achieved over the last five years.
When we set out with SQL Server in ’93, the main attractiveness to us was its downward scalability.And what I mean by that is our department heads liked that we could implement it on a little, inexpensive server, to accommodate a tight budget, and then as their business needs changed, or the business grew, we could move it to a larger server, or to a cluster of servers, as we needed to.In the upward scalability dimension, we weren’t particularly concerned, because the Microsoft product capacity has been doubling about every two years, as the hardware improved, as the software got better.And we were counting on that to handle our business growth, because it’s very seldom that you’ll double your business in less than two years.
Well, that plan worked great until recently, until the merger.
Mr. Ballmer: Until the merger.
Mr. Mayo: :Exactly.As a result of that we were facing the prospect of doubling or even tripling our transaction volume, and quadrupling the size of the database in less than a year.Now, we knew there was no way SQL Server 6.5, as good a product as it may have been, was up to that.So we started talking to the product team about the possibility of putting SQL 7 into service earlier than we might have thought.The chalk talks went really well, so our next step was to see whether the product could live up to that in the real world.
Thanks to some really great cooperation between Microsoft, Intel and Compaq, in early June we conducted a multiple-day stress test, where we ran our SAP transactions on our database, at a transaction load higher than that we were going to see after the merger.And the system held up and did the job.And that was really great news for us, because we knew then SQL 7 could do the job, it just boiled down to planning and execution, which was pretty important, because the system runs many aspects of our business 24 hours a day 7 days a week.
So to get our feet wet, and take a little lower risk, we decided to start by converting our Jiffy Lube data mart.This is a much smaller database and we had a lot more flexibility in the conversion process.So we converted it a couple of times live in the lab, and then we went to production.The production conversion took about four hours, very smooth, very uneventful.So with that behind us, we could focus in the R3R/3 conversion.After two months of meticulous planning and testing along with your team, we scheduled the conversion for Labor Day weekend.We wanted that extra day, so that we could handle any problems that might have come up.As it turned out, we didn’t need that.The system was back up operational Sunday afternoon, just a day and a half after we started.The actual database conversion took about 20 hours, the rest of the time was spent checking the conversion results, doing additional backups, doing fail safes.
As far as the results, well, the first one and the most key one is obviously the system’s been fully stable and robust for the last two-and-a-half months.That was the main criteria.But, the performance story is a little more interesting.As a result of what we’d seen in our stress test, when we announced the conversion we told our users to expect that first day the system to be a little sluggish, as the self-tuning features kind of learned the ropes and got used to working.And SQL Server 7 didn’t disappoint us at all there.The project team had been in
— we’d been in email contact with them throughout the project, and our first email that Tuesday morning when we came up to full volume was, all processors peaked at 100 percent, system flat lined, response sluggish as planned.
Mr. Ballmer: Doesn’t sound good.
Mr. Mayo: :Well, the thing is, the users were expecting it, and rather surprisingly I’d never seen a server running flat out 100 percent and still run.And so it was a quite pleasant first day, and a real good first night.
Mr. Ballmer: And from the morning to the afternoon, with all the performance improvement, that was all the auto-tuning in SQL Server kicking in.You weren’t back in tweaking that database.
Mr. Mayo: :Absolutely, everybody was under strict orders to not touch a thing, because we figured there was nothing that we knew that was going to make it any better.It was either going to work or it wasn’t.
Mr. Mayo: :So basically, where we are now, our users are really pleased with the product.They like the crisp response time and the high availability.Our DBAs like it, because of the new utilities that run a lot faster, with minimal impact on production.The SAP basis teams are really pleased at the way that Microsoft and SAP work together to create a seamless product offering.And of course, our management likes the cost effectiveness of the SQL solution.So we consider it a solid success.We finished ahead of schedule.We got more throughput than we expected.We were hoping to maintain cost effectiveness, and we actually improved it.And the system has been rock solid.In fact, in my opinion, SQL 7 runs almost as smoothly as an engine using Pennzoil.
Mr. Ballmer: Thanks.
Mr. Mayo: :Steve, before I leave I’d just like to say a word of thanks to Paul Flessner, Jean-Anne Bradley, Dwana Milton, and Jergen Thomas and the rest of the team.They put in a ton of long hours to produce this great result.It’s all paid off, and we sure do appreciate it.Thank you.
Mr. Ballmer: I want to turn now and talk a little bit about the second area of endeavor, what we call knowledge management.We’re trying to put the right information in front of people at the right time.Data warehousing, business intelligence, customer tracking, document management, these are the kinds of scenarios and applications that we encompass.Again, SQL Server 7 makes some tremendous strides forward.Where perhaps SQL Server 6.5 could target databases of up to 200 gigabytes, SQL Server 7 targets databases of terabyte.Unisys, for example, in their booth in the show is showing their new banking application, a data warehousing application for studying customer behavior and profiling.They have a database of two terabytes, with over 6.3 billion records in it.And you can go down and take a look at the demonstration and see absolutely how well that datawarehousing application performs.
We’ve integrated the Microsoft repository with SQL Server 7.So it becomes a storage point for all of the metadata that has to do with the construction of the repository.We’ve improved our storage on binary large objects, so that SQL Server 7 is a better place to store image data and a variety of other non-strictly relational data information.Also, though, we’ve done some things from a business intelligence and knowledge management standpoint which are unique.
We’ve integrated in for the first time data transformation services and OLAP services with the product.Office 2000, the next release of Microsoft Office, which will be available first part of next year, is well integrated with SQL Server 7.We’re going to have a chance to take a look at that in greater depth in our demonstration in just a minute.The way in which Excel particularly can serve as a front end to data warehouses has been dramatically enhanced, and I think you’ll find that particularly interesting.
Bill Gates, in his COMDEX keynote last night showed some of the natural language, or English query, capabilities which are built into SQL Server 7.So that literally, you type an English query, that query gets parsed and the appropriate SQL statements get generated, and the query can get executed, opening up, again, the knowledge of your organizations to many more people, who simply wouldn’t learn the kinds of techniques they would have had to learn in the past to get access to the information that they need to do their jobs.
I’d like now to show you a videotape we made with a Los Angeles Cellular, a joint company of Bell South and AT & T Wireless, operating in Los Angeles, which has already put SQL Server 7 in production in a knowledge management application.So please show the video.
Mr. Ballmer: The core of the improvements for data warehousing and knowledge management in SQL Server 7 are the integrated data transformation services and OLAP capabilities.The data transformation services essentially give us the ability to systematically and procedurally take data from any data source, from an Oracle database, even from not-relational databases, bring it together with data that you may already have in SQL Server, do any transformations – dropping fields, et cetera — that you may want to do and then export that data, as appropriate, to any destination, including the OLAP engine which is built into SQL Server.
The OLAP support comes as a result, again, of an acquisition that we made a number of years ago, where we brought in some of the absolute best people in the world, in the area of online analytical processing.And we give you the capabilities to keep the data, store it in SQL Server, to form a set of multidimensional cubes that make it very easy to navigate through and understand, summarize and analyze a large body of data; and then tools not only from Microsoft, but also from a variety of different partners in our data warehousing alliance, who are building front-end tools that work around the SQL Server 7 native data warehousing capabilities.
Again, I’d like to invite a customer who has been very active using SQL Server 7 and is in deployment, to come on out and share some of their experiences.Joining me today Lyle Anderson.Lyle is now the group CIO for News Corporation of America and was the CIO at HarperCollins Publishers when they implemented this solution.Please welcome Lyle.
Mr. Anderson: :Hello, Steve.Good to see you.
Mr. Ballmer: Good to see you.You didn’t bring your own podium, too, did you?
Mr. Anderson: :Well, I did and I got some of your favorite books here.I figured the bible of Microsoft, Dilbert.So we made sure we had lots of Dilbert books on the front here.
Mr. Ballmer: Whose bible is the Moaner’s Bench ?
Mr. Anderson: :It’s spelled well, my bible.
Mr. Ballmer: Well, great.Please tell us about your experiences.
Mr. Anderson: :Okay.Well, we’re a billion -dollar publisher of English language books around the world, and we had a pretty severe problem a couple of years ago.The industry was going downhill.We were getting returns, 50-60 percent of our books would ship out, then would come back, which is not terribly good for the profit margin.And we had no real idea of what our customers were doing.We could tell who we shipped to like a Barnes and Noble, or a Border’s or whoever, but we couldn’t tell what the end user was actually doing.So we focused a lot on getting information back in from our customers, and developed a couple of key partnerships with our customers to get point-of-sale data.And then what we did is actually created a very substantial data warehouse, initially on 6.5, then we migrated over to 7.0 last year.So it’s been a pretty exciting environment.
The solution is kind of interesting.We started out in a proprietary data warehouse about three years ago.We converted that, first of all, over to SQL Server 6.5, and our
— basically our reason for doing it was to save some money, go to a lower cost platform, et cetera.The surprising part was that we thought when we finally did the conversion over to SQL Server was that the application broke, because we were getting responses back in 5-6 minutes on a very complex query, that would have run overnight on the other application.So we were kind of going, hmm, this is interesting.And we put that in production.In about a year it grew to about 180 gigs, and now it’s up in the 2-something range, and growing every day.It’s got six years of information on all our sales.And we completed conversion over to [SQL] 7
— actually in October.And had our religious subsidiary up in September.And the same kind of problem again.
The five-six minute queries are coming back in a minute.And again, we thought we’d broke it.But just, you know, the performance improvement we’re
Mr. Ballmer: So about a five-times performance improvement on queries against the
Mr. Anderson: :Over 6.5, and probably a 10 times from 6.5 to the proprietary solution.So we were really pleased.The other part was really, you know, getting information published out to our end users around the company.I mean, it’s great to get point-of-sale data and look at it, and say okay, we’ve got great graphs and it looks pretty.But, the other part we did is we never could get the information out in a way that was really usable.
So part of this was really partnering with our data warehouse vendor, Silvon, and really building a, you know, great analysis tool, and then kind of underneath that, for the rest of the users that, you know, really couldn’t do query analysis.We built a web-based application, IIS and NT, linked it right into SQL Server.We did the first one in nine days.We had a boss that said, you will get it done.And
Mr. Ballmer: And you were making the data available widely, web format, nine days after you received the edict?
Mr. Anderson: :Yes.Fortunately, we had the infrastructure in place with SQL.But, actually publishing was nine days end to end.So a lot of this
— a lot of hair pulled out.But, it can be done.And we were publishing actually to about 200 users internally, as well as our sales force, and a selection of our other vendors that we’re working with, as well.So the end result, is right now our returns are cut down by about half.We’re a profitable company again.We’re actually
Mr. Ballmer: So this returns big.This is a big issue?This is profitability, mission critical.
Mr. Anderson: :The way this thing works, the books that go out there, three months later, after Christmas, they come back.We’d have to write them all off.And now writing them off means taking these, putting them in a dumpster, shredding them, and recycling them.So you don’t get a lot of value out of that.And we couldn’t predict what was happening, either.What we did, though, is once we got the data in and we could see what, you know, a customer coming to Barnes and Noble, like you and I every week, you know, would pick up a Dilbert.And we’d see that, we could aggregate it, we could see actually what was flowing out of their warehouses, what was flowing out of the retail outlets, and then make the right kind of print decisions, and pull back when we had to, change marketing dollars if we had to, or if it was looking like it was going to really take off, you know, bang, go in with the marketing, and really push the product.
Mr. Ballmer: And how’s your operational experience been with the SQL Server 7, you’re able to maintain good reliability?
Mr. Anderson: :It’s been the biggest un-event of HarperCollins, which is the best thing I can say.We did the conversion from 6.5 to 7.It took us about 30 hours, over a weekend.Now, we’re converting 200 gig of data, with very complex tables behind it, very complex indices.And the actual upgrade tool worked great.The other part that was really exciting was the way we did the application, we really surrounded all our old systems.And we didn’t upgrade our ERP systems, we really focused on getting information out there.So using things like DTS, and the Silvon software, you know, the combination of the two was just so powerful.
Mr. Ballmer: Pulling data in from non-SQL data?
Mr. Anderson: :From legacy systems, legacy, unstructured, file-based systems, about the worst you can have.And we were able to get it in there, we were able to get it running.And the other nice part is that we cut our batch processing windows
— I had the same situation Britt had from, you know, six hours a night down to two.And we’ve tuned it further, we’re now down to one.So we’re
— you know, 23 hour availability on data warehousing, and very proud of it.
A couple of other things I thought were, you know, interesting is just the overall performance, and the OLAP capabilities.We’re just barely breaking the surface of those now.But, you know, the power there for a user to use Excel or Access and get into the information, just pull it out, deal with it, or if you really need, you know, the high end tools like a Silvon, like a data tracker to really do the detail analysis, it’s all there.And it works.
Mr. Ballmer: Super.Well, thanks very much, Lyle.Appreciate it very much.And keep it up.
Mr. Anderson: :Thank you.
Mr. Ballmer: Good to see you.
Mr. Ballmer: I’d like to switch now and do a little bit of a live demonstration for you of some of the data warehousing and business intelligence capabilities.And we’re going to show you that in the context of Excel 2000.I’d like to invite Barry Goffe, who is product manager at Microsoft in the data warehousing area to come on stage and have a chance to share a little bit of that with you.
Mr. Ballmer: Hi, Steve.
Mr. Goffe: So, what I’d like to show you today is how Office 2000, the next version of Microsoft Office, and SQL Server 7.0 integrate really tightly to allow all our customers to be able to get access to really high-end data analysis functionality, not just customers with lots and lots of money and lots of consultants, but everybody.
Another thing that I want to show you is how SQL Server 7.0 is built on open industry standards, which allows our partners to add a lot of value to our infrastructure but, more importantly, it allows customers to have an enormous array of choices in terms of what tools they use to solve the different business problems that they have.They can choose different tools for different problems for different communities within their organization.
Mr. Ballmer: As we heard Lyle discussing at HarperCollins.
Mr. Goffe: Absolutely.And then, finally, we’ll talk a little bit about performance.We all know that SQL Server 7.0 is a breakthrough release, but let’s try to make it real and show people exactly what kind of performance they can expect.
Mr. Ballmer: Super.
Mr. Goffe: So, to start out with, what I want to do is show you Excel 2000, which again is the next version of Microsoft Excel, and in particular I’d like to take a look at some pivot tables.And some of you may have used pivot tables in the past, but the important thing about pivot tables, it allows you to basically slice and dice your data.
Mr. Ballmer: The pivot tables I use today only let me slice and dice really through the data that fits in the memory on my little machine.
Mr. Goffe: Well, that’s exactly the point.In the past, with pivot tables, you were doing all the processing on your machine, so you were limited to the amount of data that you had access to.And also, to get the data to the pivot table initially, I mean, you had to be a database programmer, you had to go out and issue queries and get the data out of your databases and onto your laptop.Well, no longer do you have to do that, because Excel 2000 is running directly off of SQL Server 7.0’s OLAP services.
So, what I can do here is, first of all, it’s really easy.Excel 2000 is taking a look at what kind of data do we have in SQL Server 7.0, and I can simply drag and drop to build my pivot table.So, for instance, if you’re talking about sales information, let’s say drag and drop geography and time and maybe our actual year-to-date sales.And we can drill into that information.So, for instance, we can take a look at Europe sales, and maybe drill down into USA sales.
Mr. Ballmer: And this is just Excel’s standard capabilities front-ending an OLAP database in 7.0?
Mr. Goffe: That’s right.What we’re doing is, we’re navigating through the hierarchy that’s defined in SQL Server 7.0 that basically describes our data.
Mr. Ballmer: Excel just knows about that because of the integration.
Mr. Goffe: That’s right.And I can continue to drag and drop here.I can look, say we’re an insurance company, I can look at my sales by product line, I can throw on my forecasted sales, and take a look at how those forecasted sales compare to our actual sales.
Okay, so now we’ve done a little bit of analysis with Excel.Let’s move over here to our other machine, and let’s take a look at what it takes to build a data warehouse.Now, traditionally, one of the most difficult parts of building our data warehouse is getting all the data into it in the first place.Most companies have data in lots of different systems.I could have my sales order processing system on a mainframe.I might have my accounting system on NT.I might happen to have a manufacturing system that’s running on UNIX.So, how do I go out and get that data?
SQL Server 7.0 is the first time anyone is bundling a tool for moving and transforming data.So, this comes as a standard part of SQL Server 7.0, which is a unique thing in SQL.And what we’re looking at here is, data transformation services that have a wizard that allows us to go out and get that data.So, what I’m going to do here is, let’s just say I have some data out in an Oracle database.
Mr. Ballmer: A lot of our customers do.
Mr. Goffe: And I want to go pull that data in.So, the first thing that I’m going to do is, I’m going to go log-in to my Oracle database.
Mr. Ballmer: So, not only can we take data in from Oracle, we can systematically allow you to keep your transactional application in Oracle, but still take advantage of and build rich data warehouses with SQL Server?
Mr. Goffe: Absolutely.We’re fully interoperable with any platform, not just with Oracle.I can get at any kind of heterogeneous data on any platform.Okay, so now what it’s done is, it’s gone out to Oracle and shown me, what are the tables that I have security access to.So, let’s just go ahead and pick the first table here.
And what’s interesting is, not only can I move the data from Oracle to SQL Server, but I can also transform that data.I can validate the data.I can summarize data.I can do all sorts of things with the data.In this particular case, let’s just go ahead and move it.
Well, it’s finished here, so it’s going out to Oracle, reading the data, populating a brand new table in SQL Server, and there it is.It’s pretty fast.We just moved 10,000 rows from our Oracle Server into SQL Server.
Mr. Ballmer: Wow.Great.
Mr. Goffe: So now what I want to do is, do some analysis on this data warehouse that we just added some data into.And, again, SQL Server is built based on open industry standards.So, what that means is that I can use Microsoft Excel to do my data analysis, or I could use any of the three dozen tools from third party vendors that are going to ship in the next 90 days that also support SQL Server 7 and OLAP services, or we’ve also done work to integrate Microsoft Visual Studio in with SQL Server 7.0.So, I could build my own.And, in fact, that’s what we’ve done here.Using Visual Studio, we’ve built a pretty advanced data analysis tool that allows me to visualize some interesting things about my data.
Mr. Ballmer: How long did it take you to build this thing?
Mr. Goffe: It took us a couple of days to throw this thing together.So, the interesting thing here is, what I can look at is how different parts of my business are reflected in other parts of my business.So, for instance, remember we’re an insurance company, we can look at sales by product line in the top part here, and sales by quarter in the bottom part.
Now, wouldn’t it be great if you could say, show me how much commercial product I’m selling on a quarterly basis.
Mr. Ballmer: Yes, show me that.
Mr. Goffe: Okay.Well, it’s basically just a click.I just click on commercial.And now what I’m seeing in the bottom is how much business I did on a quarter by quarter basis for that particular product line.
Okay, now maybe we want to do something a little bit more advanced.We want to start analyzing our budget.Okay, well, now we can look at even more information simultaneously, and figure out how everything affects everything else.So, now on the top, we’re continuing to look at our sales by product line and sales by quarter.On the bottom what we’re looking at is how much we’re budgeting for advertising for each of the different media types, as well as how much we’re spending on advertising on a quarter by quarter basis.
Mr. Ballmer: All right.
Mr. Goffe: Now, what’s really exciting here is, I can make changes to the data and do what-if analysis.Now, this is another unique feature of SQL Server 7.0, the ability to do online what-if analysis, and see the changes instantly that are occurring back in the database.
So, for instance here, I can take a look at my telemarketing budget, and you’ll see that, hey, it’s pretty low on a quarter by quarter basis.What would happen if I increased the amount of money that I’m spending on telemarketing?Let’s go ahead and do that.Let’s just drag this little bar here, and increase the budget for telemarketing.And what you’ll see is, we’re getting some pretty dramatic increases.In fact, in the liability product line, we’re getting incredible increases.
So, with this model and with this what-if analysis, I’m able to do some really powerful
I get a really powerful understanding of how my business works.
Mr. Ballmer: Super.
Mr. Goffe: Now, another really exciting feature here is that SQL Server 7.0 allows me to take a snapshot of my data, and take it offline to do analysis when I’m not connected to the server.
Mr. Ballmer: So, a guy like me, who mostly uses his computer on the airplane, you’re saying I could take with me this essentially little data mart and do this analysis and what-if on the road?
Mr. Goffe: Absolutely.In fact you don’t have to have the server on your laptop.We just send you the data on your laptop, and you could use Excel to do this analysis, or if you need to, use third party tools.So, in fact, we can click down here, and the first thing it’s going to ask me, hey, do you want to save those changes that you’ve made to the database?And we’ll say yes.
Mr. Ballmer: You can save them back into the database?
Mr. Goffe: That’s right.So, we did this what-if analysis, and now we’re updating our multidimensional store with these changes that we’ve made.And the next thing that it’s going to do is, it’s going to go ahead and create a snapshot for offline analysis, and embed that in an Exchange email.
So, here we go, here’s our Exchange email.I’m going to go ahead and send this to you, Steve, so now tomorrow or Wednesday, when you fly home to Seattle on the airplane, you’ll be able to do this analysis.
Mr. Ballmer: Great.Great.
Mr. Goffe: So, what we’ve seen is, how, you know, we’ve built SQL Server with open industry standards, and it gives our customers an enormous array of choices in terms of what products they use to solve the different business problems that they have.
Mr. Ballmer: Is it fast?
Mr. Goffe: Absolutely it’s fast.
Mr. Ballmer: Now, Lyle said it was fast, but is it fast?
Mr. Goffe: Well, let’s show you exactly how fast it is.What we’ve done here is, we’ve built a little data warehouse that models kind of a medium-sized grocery chain.We have about 10,000 customers.We have a couple thousand products.Most importantly, we have 150 million sales records in our data warehouse here.Now, let’s send a pretty complex query to this data warehouse and see what kind of performance we’re going to get.And the query that we’re going to send is not a trivial query, but it is exactly the kind of questions that you would probably ask of your business people.Hey, I want to know this information.So, the question that we’re asking of the database is, show me which ones of my stores are growing their sundries business, their non-consumables business, faster than their food products business.And analyze that on quarterly sales.
Mr. Ballmer: Yes.
Mr. Goffe: Let’s go ahead and click on show me the sales, and hey, that was pretty much instantaneous.
Mr. Ballmer: You’re saying it just queried 150 million records and came back that quickly?
Mr. Goffe: That’s right.And it was not a trivial query.It’s a very complex query.Now, you may think that, hey, this webpage was cached.
Mr. Ballmer: I might think that.
Mr. Goffe: So, let’s go ahead and just prove that it wasn’t, and I’ll just click on the refresh button here, and you’ll see, again, instantaneous.Basically, it’s about one-tenth of a second to issue that query and retrieve back the result set.
Now, the most important thing here is that up to very recently, it used to take huge, huge hardware to be able to get that kind of performance.In fact, you know, million dollar servers would give you that kind of performance.And this demonstration that we’ve put together is running on a server that cost us exactly $30,200.
Mr. Ballmer: $30,200?
Mr. Goffe: $30,200.
Mr. Ballmer: Do you think anybody else can do this kind of stuff on a machine that costs $30,200?
Mr. Goffe: Absolutely.Absolutely.Anyone can be able to do this.
Mr. Ballmer: Anyone can do it with our product.
Mr. Goffe: With SQL Server 7.0, absolutely.
Mr. Ballmer: Well, that’s great.Thanks very much, Barry.
Mr. Goffe: Thank you.
Mr. Ballmer: I want to turn to the third leg of what people need in business, and that’s this area of electronic commerce, whether that’s the reengineering and supply chains or corporate purchasing, or direct selling and support, there’s a lot of work, again, in SQL Server 7.
The way SQL Server 7 integrates with Windows NT and its Internet Information Server, we’ve enhanced to optimize the performance of people who want to put a web front-end on their SQL Server 7 based application.SQL Server 7 supports full text searching against the database, so that you can literally expose the information that you include or contain in a SQL Server 7 database to anybody you want to who’s coming in on your website, and they can do a traditional web-style English language query, and all of the data that happens to be actually stored in the database can be similarly full-text indexed and made available.
We needed a platform that is complete, so that it is cost-effective, to actually to put together the kinds of complicated applications that people want to do, and of course this is where scalability is really required, because in a way you’re inviting your business partners and your business intelligence applications.
Let’s hear for a minute from CBS Sportsline, one of the largest providers of sports information on the Internet here in the United States about the ways in which they’re using SQL Server 7 for electronic commerce.
Mr. Ballmer: The kind of experience that CBS Sportsline had is largely the result of the fact that SQL Server is part of a platform of products for building Internet-commerce-style applications.It takes advantage of and works well with the Internet information services, the transaction services, and the queuing services that form the backbone of our application server strategy on Windows NT Server.It works well with our development tools.It works tightly with our Site Server product, which provides a range of advanced business-to-business and business-to-consumer e-commerce facilities.And it is open and available not only to web client applications, which is very important on the Internet, but also to clients running Microsoft Office for the richest authoring and analysis in the case of the intranet.
I’d like now to invite to join me on stage Alan BarapasBourassa.He’s the director of strategic systems for barnesandnoble.com.Barnes and Noble is the largest bookseller in the world.Obviously, a lot of interesting issues for them right now in terms of the Internet.And Alan is going to tell us a little bit about their experience.
So, please welcome Alan.
Mr. BARAPASBOURASSA: It’s nice to be here today, Steve.
Mr. Ballmer: Well, it’s good to have you.Tell us a little bit about Barnes and Noble and maybe some of the
hey, nice picture.My kids would like it anyway.
Mr. BARAPASBOURASSA: We might have to get exclusive author rights on you.
Mr. Ballmer: Tell us a little bit about Barnes and Noble and some of your challenges.
Mr. BARAPASBOURASSA: Today, Barnes and Noble has over a thousand bookstores nationwide with 1997 annual revenues of $2.8 billion.The company’s website, barnesandnoble.com, is listed as one of the 25 fastest growing websites, and one of the top five commerce sites in the world today.Projected sales for our web business are $60 million this year.
While we are the largest bookstore retailer today, we were slightly behind the competition to build a business on the Internet.Our biggest issue was that while we had built an online presence, our back-end systems were batch-oriented.And we needed to move into more online real-time processing to improve our order fulfillment, customer service, and plant floor operations.We wanted to cut down on end-to-end processing time to be able to maintain responsiveness for our online customers as our volumes increased.Basically, we needed to build something quickly, but also make sure that it offered the quality buying experience our customers have come to expect in our stores.
This meant building a new state-of-the-art fulfillment and service system available around the clock.
Mr. Ballmer: And what did you do?How did you meet these challenges?
Mr. BARAPASBOURASSA: Well, we decided to enhance the back-end distribution system by transitioning from our existing system to an online real-time SQL Server 7.0-based shipping, order management and financial reporting system.We built our BackOffice system using a three-tiered architecture with key Microsoft components, including Visual Studio 6.0, Microsoft IIS, Transaction Server, Cluster Server, Message Queue Server, and of course SQL 7.0 — it was the entire suite.
High availability and reliability are also critical to our operation.Microsoft Cluster Server provided us with immediate and automatic failover of our back service operations in the event of a systems failure.Flexibility is also mandatory.Microsoft Transaction Server and DCOM allowed us to reconfigure and redistribute applications services to respond to sudden changes in processing requirements.
Mr. Ballmer: How about the result?Good, bad?I know this is mission critical right now, I certainly know, for the top of Barnes and Noble on down.
Mr. BARAPASBOURASSA: And the results have been outstanding for us.We’re really, really pleased with our implementation.And by using SQL Server 7.0 to scale to terabyte-sized multiprocessor systems, we will be able to configure our combined database over a terabyte.In production, we have seen more than 100 percent performance improvement to date.We are even expecting to squeeze more out of our system.
We have a new program called Million Title Express.If a customer in a Barnes and Noble store can’t find the book they want on the shelf, it can be ordered in the store utilizing the industry’s largest online inventory, and immediately shipped to the customer’s home.That was one of the advantages actually we got from this online implementation using our new products.
The elimination of batch processing makes our shipping on the plant floor flow much smoother and faster 24 hours a day, seven days a week.This has improved the speed of delivery of all of our books to our online in-store customers.It also enabled us to satisfy changes to customer orders up until the moment the product is actually shipped out of plant.
Having implemented this new BackOffice system has given me an opportunity to work with a great group of people, and barnesandnoble.com is now poised for the future in the e-commerce arena.
Mr. Ballmer: I find it absolutely tremendous, and as a great Microsoft customer, and as somebody with a very competitive struggle, we’ll support you our best, and we wish you the best.
Thank you, Alan.
Mr. BARAPASBOURASSA: Thank you.
Mr. Ballmer: Okay.We’ve had a chance to talk about the three areas in which SQL Server 7 will support the aspirations of the business.But these applications need to be built and managed by a set of developers, database administrators, and IT professionals, and it’s essential that SQL Server 7 serve their needs.
Again, we’ve made a lot of great progress.The universal data access in SQL Server 7 allows integration with and access through OLE-DB to a variety of different data sources.Our support for Unicode is, I think, unparalleled in a way that allows people like Barnes and Noble and others to develop applications to serve users around the world with the same application.We provide a set of new debugging and profiling tools which, again, are designed to make it easier and quicker to build these applications.In a way that’s fairly unique, SQL Server 7 can actually be embedded in an independent software vendor’s application, reducing the time and complexity of installing the application.We’ve added visual design tools, and perhaps in a way most significantly for developers, we’ve added a mobile version of SQL Server 7.
Many of the most popular applications that people are building today, whether they are business intelligence applications or transactional applications, require a user to go offline, for a sales rep to go offline to take orders, and then want to resynchronize with the rest of the business.The merge replication facilities that we provide, we think are unparalleled.SQL Server 7 in the mobile form can run in as little as five megabytes of memory, and it joins a family of offerings through our enterprise version of SQL Server 7 that lets us target the individual user, small and medium business, the branch, the department on up to the enterprise with the same code, 100 percent compatible applications from a developer’s perspective up and down the line.Again, very important for the mobile scenario, the branch office scenario, et cetera.
From the database administration and IT professional standpoint, we also had to make great strides.We talked through the session about our support for much larger databases.I’m going to describe in a little bit some of the work that we’ve done to improve our ability to perform online maintenance.We’ve built in tools that help you automatically monitor and schedule operations against the database.And in a fairly unique way, we’ve provided graphical tools for query analysis, as well as the tools to manage large numbers of SQL Server 7 databases all together, particularly important in environments like a branch banking or a retail operations condition.
I’d say from the standpoint of the IT professional, the most important issues are around reliability and availability, particularly for very large databases.For those of you who have not had a chance to check it out, I’d encourage you to go to microsoft.com and look at our Terra Server application. It’s well over a terabyte of geographical information with aerial photos of literally the whole world.This is an application that has been in production seven days a week, 24 hours a day on microsoft.com for the last three months.
I mentioned the Unisys application, which you can see on the show floor.Data General has a new terabyte-plus clinical management application for health care professionals, in which they store X-ray data, scan data, and other patient record information in a large database.Data General is providing a service that guarantees 99.9 percent uptime on SQL Server databases.
With Compaq, we’ve worked on this issue of how do you keep the database up and running while you’re maintaining it.Backup is a big part of the problem.You’ve heard from some of the earlier speakers about the improvement.We now can do a peak of about 602 gigabytes per hour of backup in SQL Server 7.And even if you add to the transactional load against that database, we see no more than a 5 percent performance degradation in the speed of all operations.This is the kind of architectural support we need not only in terms of quality, but architectural support to make sure that you can do multiple things against the database and keep it up and running seven days a week, 24 hours a day.
That’s what we’ve done for IT people, for developers to help them satisfy the needs of the entire user community.But to really make this product successful, we needed to do more.We needed to make sure that the independent software vendor community was behind it.There will be over 300 applications in 90 days, and 3,000 in the next 18 months.SQL Server 7 itself will become available through our resellers in January, and then these applications will follow.If you look at the big five accounting firms, Price Waterhouse, Coopers, KPMG, Ernst and Young, Arthur Andersen, Deloitte and Touche, every one of them now have application practices, ERP, e-commerce, sales force automation, in which they will design in and install SQL Server 7 as part of the line-of-business applications that they engineer for their customers.
We’ve invested in the training of over 52,000 database administrators.We’ve got partners like Cambridge Technology Partners, Great Plains Software serving medium-size enterprises with their ERP systems, J.D. Edwards, and many, many others have done really phenomenal work.When I stop and see some of the kinds of things that, say, Great Plains or Baan can demonstrate, the kind of accessibility to information, the kind of rich e-commerce scenarios that people are supporting on SQL Server 7 are really amazing, and I think very, very important.
The most important thing with all of this great new capability, with all of the kinds of things we needed to do to support these applications, though, was to make sure, 100 percent, that SQL Server 7 was rock solid.At Microsoft, we run now entirely on SQL Server 7.Microsoft.com, our SAP installation, our data warehouses, we’re in production at over 10 customers, we’ve done conversions with customers of over 1,000 databases with 98.5 percent of them converting with no additional work.
We have over 100,000 beta sites done and in production.We needed to make sure that the mobile users would really see an environment where every end-user did not need to become a database administrator, done.But perhaps most importantly, we wanted the major ERP vendors to really say this product was ready, that it had the scalability to target 95 percent of their customers, that it had the quality, the reliability that was necessary to take this product and make it an important part of the solutions that they deploy in the biggest enterprises.
So, we told our team we’re not shipping until we meet the bar.And so, even now, as we finish up, and we’ve gotten what we think are the final bits, we’re going back again.We’re putting the final bits in production in Microsoft, and a number of other customers.We’ll run that way for a while before we’re finally free and then make the product available through resellers in January.And certainly we think we’ve done what you all expect us to do in terms of focusing in on quality and reliability in exactly the right way.
As I said at the start, SQL Server 7 is a breakthrough product.Scalability, performance, availability, e-commerce, Office integration, ease of use, data warehousing, OLAP capabilities, the partners, the applications, it is a significant product that I think allows the PC, not just Microsoft, but the PC to really assume a different place, an even more important place, in the enterprise, in the small business, in the e-commerce realm, a very, very important place.
I’d like to bring on stage now essentially the leaders, at least the leaders who are here in Las Vegas, of our SQL Server 7 development team.I want you to look at their faces.
Come on, come on on board fellows.
Mr. Ballmer: You can go ahead and applaud for them, too.Now, there’s a reason why we brought them on stage, besides to give you the chance to recognize them.We’re having a party tonight, 9:00 in this room, I want these faces recognizable for you tonight at the party, for them it’s a chance to do real-time selling and technical support.So bring your questions, anybody here in the group.I’ll actually give you a little special bonus, these two guys right at the end are really the top two managers on the project, Paul Flessner, [email protected], and David Vaskevitch, vice president in charge of this area, who really built the team, [email protected] they want to know your questions, your thoughts, your problems.Feel free to exercise them regularly, and please join me in not only thanking these guys, but we have a group, I think over the satellite, joining us from Redmond, Washington, the rest of the SQL Server 7 team.
It’s been our pleasure being with you today.And from all of us, and do we have our team in Redmond, or did we lose the link.We say thank you all very much.We appreciate your time.
Mr. Ballmer: Thank you all.We’ll see you tonight.
(End of event.)