WinHec ’98


March 25, 1998, Orlando, Florida

[Due to the varying sound quality and subject matter of tapes, the information in this transcript may contain inaccuracies.]

MR. Allchin: Good morning.I just got a chance to look at some of the survey results that some of you filled out.There are some pretty amazing things in here.About 50 percent of you have never been to WinHec before.Welcome.

There’s also a smaller percentage of you who have been to all seven.Welcome and welcome back.

Some of the other questions I thought were interesting — memory size.What’s the standard PC memory size in 1998, by the beginning of 98?Sixty percent said 64 megabytes.We like you.Ten percent said 128 megabytes.We really like you.There’s 20 percent that said 32 megabytes.By the end of the conference, hopefully we’ll convince you that it should be more than that.

The last question that I’m going to share with you was, in rank and priority of order, what do you think are the most important challenges to the PC industry and improving customer experience with PCs?

The number one issue was device-driver quality capabilities.It’s something that together we’re really going to have to Work on.

The number two issue was just advancing the PC system architecture in general, and, obviously, this whole conference is about doing that together.

I’d like to talk just a few minutes, briefly touch on three things.There’s the Windows part where — (inaudible) — to the industry, the key challenges that we see for Microsoft and which we have to Work on together and what we want to do about it, and then a roadmap both of Windows and then a little bit about our future plans in terms of technology.

The industry is doing incredibly well.In the days of 1992, there were 25 million PCs shipped.This year, we expect about 90 million.So the PC has continued to morph and grow and expand, and the amazing PC has gone all the way from hand-held, embedded devices to the largest server clusters.Meanwhile you’ve been challenging us to stay up with you, and we have tried to create operating systems and a platform to try to show off the best of that.

And on this big screen, you will see how we’ve tried to map our operating system onto this amazing morphing that’s been going on.We do have an incredibly strong following in terms of Win 32.This is not Windows.It is Win 32, a programming interface for Windows 95 and NT.As of November 1997, we’re up to about 90 percent — actually, the newest figures will show it’s over 90 percent — of developers who are focusing on the Windows platform.

So in an era where you might think that there are challenges that are being mounted to this, in reality the developers are telling that they are locked on to the Windows platforms.

Another thing that we see happening in the industry is the amazing movement of Windows NT workstations.This is IDC’s data in millions of units from 1996 to 2001.So if you look at 1998, you’ll see that they were projecting 8.4.This survey was done last year.Even though we’re only into the third month of 1998, we know that this number could be off by a lot.We could do about double this number.

So what are the challenges that we can see?Well, there are really three of them.First, we need to improve the customer’s — (audio break) —

I think we need to lower the cost overall platforms, the total cost of operations.We need to move into the enterprise.

Last, there are a series of new, interesting areas that we can leverage together and not only scale up and down, but we can move to the side and open new opportunities for both of us.

First, the challenge of improving what the customer sees — the interface.We build computing systems.That’s true in the software and true in the hardware.The number of questions that we get on our support lines imply that we together haven’t done a very good job.The questions I get from my mother imply we haven’t done a very good job.

Systems don’t function out of the box.We actually did a few surveys where we went and bought systems, and we were surprised at how many, you plug them in and they actually didn’t work.That’s going to give us all a very bad reputation.

If you upgrade the software or hardware, it’s way too difficult.For as many times as we talk about how great plug-and-play is, well, I replaced my tape drive the other day in my home machine, and, plug-and-play worked, but I also cut my hand and had blood on my hand because I cut it on the sheet metal, and that’s not very good.We can’t expect this to continue to grow, the industry, with that kind of interface.

Why?A lot of it is the operating system, the interface — (inaudible) –we add anything to this, new software or hardware that could become stabilized.We allow nonfunctional or poor-quality drivers — that’s a message you’re going to hear over and over during this conference, and it’s something that we together are going to have to do something about.

Lack of adhering to specifications — that’s also true for both of us — and an undetermined existing system of technology.Do we really know what is — (inaudible) — system?Is there really a way for the software to pick it up and do the right thing with it?

From our perspective, we want to focus on simplicity and quality.And in the simplicity area, together we’ve worked on plug-and-play and it’s come a long way, but we need to move to externally connected devices.The 13 and 94 you have seen are just examples.People shouldn’t be cutting their hand on sheet metal.

Power management support for portables.Well, we’ve tried once.We’re trying again.I’m actually incredibly excited about all the work that’s going on in this area.

Improving the user interface is crucial.There need to be less error messages, and the ones that do appear need to be understandable.We have quite a few in Microsoft that we’re now in the process of looking through, and we need to figure out what we can do to increase the amount of auto-correcting in the information that’s coming to the user.I mean, asking somebody, you know, where their only response is OK doesn’t really make any sense.

The last thing that we’re focusing on is our Wizard technology in helping in making network connections and adding hardware and the like.

In terms of quality, we’re committed to improve our system testing.Now, we do this for code coverage analysis.As we test a system, we do it to make sure that we get as much code coverage as we possibly can — (inaudible) — we do it for mapping stress on the system, but we still have a way to go.Obviously, the fact that we have to do service packs with the many bugs that are in the service packs that we’re fixing, we find we have to do a better job.

We also have to continue to improve our hardware compatibility test.We have gotten a lot of feedback on this, and that’s something that we’ve put a very senior person in charge in that area and they are working hard to improve the process, not only the tests, to have better samples for you, and making sure that together in the hardware guidance that we are focused on making — ensuring that there’s a quality experience for that end user.

Driver signage is something that we’re in the process of moving into.It won’t be turned on by default, but you can expect in the future that it will be one way for us to assure that it’s been a thorough test pass.And then self- resolving errors, is there something that the system can figure out what the problem is and move on, even if the error still exists?We need to do that.That’s true on the — (inaudible) — level, where the bad memory block shows up and we can work around it or even in the computer space.

Well, we’ve got some cool demos.One of the ones that we want to — (inaudible) — NT5 is some of the portability capability that we were doing there.Our goal is to dramatically improve the mobile experience.We’re going to do a demonstration showing standby — that’s a way of hot-docking battery management, hot-floppying most of the floppy and the battery in hibernation.

Patrick Franklin from the NT group is here, and — (off-mike conversation) — hi, Patrick!

Patrick Franklin conducts demonstration of Windows NT 5 plug and play and power management features.

MR. ALLCHIN: So what do we think we need to do together?We need to simplify the software and hardware.I’d say the key there is external accessibility.Plug-and-play’s obviously a critical part of that, so we can do things like was just shown here with the hot docking.There’s also automatic operation and discoverability where we ensure that if somebody does an outside operation with a device, like slipping in a piece of paper in a scanner, the system just detects it and does the right thing, and, from the discoverability side, that if you add something — again, even if it’s not paper or the like, you add another device, it can just sense it.

Again, focus on the quality of the software that you write, as well as your hardware.We certainly will try to help in terms of our driver samples and our process and our technique, and together we need to follow the specifications as closely as we possibly can.

Our second challenge is reducing for the overall platform total cost of operation.What is the challenge?This is no surprise to you.This is a quote from Forrester:
“PCs are disorderly corporate citizens, expensive to support and prone to disaster.”
That doesn’t bode well for the industry.
“Unless large companies can manage the desktops, cost and risk will accelerate a client-server scale.MIS must begin to standardize and leverage desktop management tools to get this beast under control.”

So what are we doing?We have a goal to reduce the TCO by half with Windows NT 5.Most of the data is from the Gartner Group, looking at technology improvements that we’ve made through the years — let’s take the Windows terminal support.Our view is that people want more than just terminals, in most cases.They really want the ability to have storage — an example here of the laptop, being able to move those around, work on the plane, come back in and get your information.

We’ll be able to do almost all of that with Windows NT 5 to a degree that has never been done before.

In particular, we have this initiative called the Zero Administration for Windows initiative, and the goal here isto get to that 50 percent drop in cost and keeping that up, and we’re doing this in a variety of ways.First, the DLL help problem of replacing random DLLs and not being able to tell the interfaces or the relationship between the DLL you are trying to fix.Now, we will have a capability that has a miniature database, if you will, that remembers all those relationships.

And the way we’re going to encourage software developers to write their applications uses the technology that will feed into that data base, so we know the relationships.

And also a series of things we’ll be able to do — centralize the penetration of that so once you do that, you can install these databases on the network and the right files will come down and will be used and it will understand the relationships, whether part of it is on the client or part of it’s on the server. The second thing we’re doing is to really focus on OS management and deployment issues.So the key thing is that once there is a service pack for a server OS, it can be deployed across the enterprise.

And when the systems boot, if it’s the operating system, the first time it boots, it’ll try to figure out through the directory service what operating system are they supposed to have, and it’ll automatically download it.This is going to be true whether there’s a boot on RAM in the system or not.Obviously, it would be nice if it’s right there in the LAN environment.

Now, if the operating system has the right version on the disk, it’s not going to talk on the network anymore.It’s just going to go figure out and tell the right — (inaudible) — to just use it.So an important concept in the operating system is the cache on your local disk.

Another thing we’re doing is we’re enabling split streams to happen so that as we come out with service packs and changes to the system, we can integrate those into the original disks, so you don’t have to go through this painful, lay down the operators and then put down the service pack.Very cool technology. Actually, you may see some of this later on in the week.

Another key piece of technology that we’re using to attack this problem is Intellimirroring, and there’s almost two levels of Intellimirroring.The first one is just managing user documents that may be the olderdocuments on the desktop.My documents on the desktop may actually not be physically for there — physically for there in the sense that it’s a cache, but the real copy was out on the network someplace, and it’s automatically synchronizing — (inaudible) — and it’s also true of the preferences things like Internet favorites or customization into the desktop or your applications — this will all be gotten when you log in through the directory server.

You can take Intellimirroring to a whole new level, if you want, where we create a stateless (ph) client so that everything on the client can be carried.That, of course, has additional resource environments that are necessary on the server.If the information is local, we’ll use it on the local system, but if you change it, it has to be flushed back.

Now, the reason for the full Intellimirroring is so that we can do machine replacement with the current state of our software if you want.So as we move to a more structured state, where we keep it on the system, you probably won’t have to go to this level, but today if you want to do a replaceable machine environment, you have to go through the full Intellimirroring.I think at Bill’s talk tomorrow, there should be a demonstration of that.We’ll take a machine that actually — that has some problems and take a new one, something that’s completely discoverable, and have the operator just load it and go right back to the space on the system the way it was.

The last area is using the directory service and the Windows-based enterprise management for web-based enterprise management technology, we’ll be able to control the system better in a virutalized locale as well as be able to have the instrumentation that the operating system can leverage.

We have spent a lot of time on this total cost of ownership problem.There’s a cost of ownership problem for any of our partners that are loading the system, as well, not just for end users.But also, we’ve gotten a lot of feedback from you about the user-install process — what it’s like coming straight to the interactive — (inaudible) — and they think it’s a boot.We’ve really improved that.We also wanted to make tools better for diagnosing problems.In Windows NT, there really wasn’t any safe mode for the system at all, and we also didn’t have any way to really put the system back with a disaster recovery backup and restore.

We’ve added most of that.It will be able to fit almost in a mini DOS form — a command line, actually, not DOS form.Then you’ll be able to do it in a secure way.

The last, which I think you’re definitely interested in, is simple deployment tools for both corporate, but especially for our OEM partners so they can do disks very fast.We’ve come up with this technology, with great feedback from you in helping design it, called Disk Clone.Now the idea of Disk Clone is that you can make — set up your configurations, map it, if you will, onto disk, and then just hit copy to this thing and plug it into the system.And when the end user boots the first time, they’ll be exactly back to the way you configured it, with a lot less overhead than there typically used to be.

Valerie See is here to give us a demonstration of that technology.Hey, Valerie.

Valerie See conducts Disk Clone demonstration.

MR. ALLCHIN: Take it straight.Thank you.

If you ever saw that your team forced setups, you’ll appreciate the advantages here, because it was pretty bad before, and I think we’ve gone beyond what previous cloning technology has offered in terms of disks, too, because there’s a lot more ways that you can configure the system. Then you just basically snap it and roll it back to exactly the place that Valerie showed.

So, together, what do we think we need to do?We need to utilize improved setup and install technology.We’re not quite done yet with the CDs we’re giving out here for NEW that you’ll get later on this week.You’ll see that the setup process itself, the user interface, we’re just not done with that yet.But you will get an idea, and Carl will talk about what we’re doing, but we certainly wanted to take advantage of all this cloning technology.

We also want you to write WMI drivers and we want you to take advantage of the welcome technology if you’re trying to get instrumentation data presented in the system.And also from the corporate perspective there are things that you could do to help us in the Iintellimirror area, and there’ll be a separate talk on that program–(inaudible)–but also in terms of intelli-mirroring and the directory service.

The last challenge is broadening the PC into new user scenarios.Now, at first you could say, “Is this really a challenge or is it an opportunity?”We actually think it is an opportunity.We’re even scaling the PC environment up and down and we’ve still got a long ways to go.We still haven’t really attacked the super high end, and there’s all sorts of little devices that together we haven’t moved the PC industry into.We’re getting there.

It’s not quite enough to use what we currently have.We have to be inventive about new technologies, about ways of leveraging the powers of PC–home automation.There’s no reason why we can’t move into that space and offer–the PC industry offer a solution in that area.

So what’s the market potential?We don’t know.I can tell you that we’re investing in both moving the PC technology up and down, as well as finding other areas, whether it’s–(inaudible)–home automation, or whatever.We’re looking for ways to have the PC industry participate in that.

In terms of scaling to the enterprise, this is the ability–(inaudible)–and I know if you’ve ever dealt with any enterprise customer, these are the words they say:”The PC industry still has a long way to go to answer what they want in these areas.”Obviously, I’m not going to drill into the whole of things we need to do, but reliability is at the top of the list.That’s something if we really want to crack the enterprise–(inaudible)–ability there as well, and then scaling at the bottom is

something that I’m going to talk a little bit more at the end of this talk about.It’s certainly something that looks very keen.

We’re doing a bunch of things to scale up in Windows NT 5.Storage–whether it’s hierarchical storage management, whether it’s jukebox support–(inaudible)–or other types ofoff-line media.We’ve added a lot of technology in that area, and we’re working with partners.We’ve come a long way to being able to crack that high end.

Large memory support–there’ll be very large memory addresses, 64 bit addresses for the alpha–(inaudible)–improved FNCs–(inaudible)–and we will scale in terms of the number of profits, making even higher numbers. We continue to look for ways in which other bottlenecks–(inaudible)–don’t get in our way.Today memory management is a big problem in the PC industry for us to be able to attack.Certainly the–(inaudible)–will help and the chips in there, but it’s–(inaudible)–

There’s also scaling down to the home, taking our technology and enabling us to stretch ourselves in this scenario.There’s a whole bunch of stuff together we’ve got to do.If you’re using a 28.8 modem, it’s hard on the Internet today.That’s just got to change, and we’re working with you and everyone we can find, to be able to gain higher performance in terms of whether it’s your DSL or it’s your cable modem or the world today and why you can’t get that high-speed environment to come.

If you’ve ever been on an ADSL trial, it’s amazing.It changes your life, the way you think about PCs.Instead of going to the newspaper and looking up what the movies are, if your PC is always on, using the technology that you see here, and you’ve got new information readily available–6 megabits per second or even less than that–that it’s always right there, you can always find what happened to your news, what’s happening in your town, where to go.

And every trial that we’ve seen, it has changed the way people think about PCs.We’re going to sell more PCs the more we push the communications edge.

We also haven’t moved into the entertainment space as much as we should or could, and the family uses the PC–(inaudible)–in order to do this, we have to make the PC more compelling for a broader range of users at home.

Now, one demonstration that we want to talk about today is how we can have more families use the PC, both for work and for play, if you will, and one scenario is to use the ability of multimonitors, where you can work on your PC at the same time doing–(inaudible)–on one of the other monitors.

It allows the PC to sort of become associated with the TV, which is good.We can do this in a variety of different ways using co-ax or wireless–(inaudible)–and, obviously, it’s not just the software–(inaudible)–, we’re doing it with you, and there’s lots of ways to do this with the hardware.

Dave Marshall’s here to give us a demonstration of this, and I think on the screens you can see the general idea–well, you could.They’re resisting.Oh, well.

Hi, Dave.

Dave Marshall conducts demonstration

MR. ALLCHIN: So together, we can expand product families, we can move up into highest server range.We can enable new uses.We haven’t shown any demonstrations of what could be possible using standard operating system technology, standard PC technology — things like switches and dialing systems, home automation and other kinds of embedded devices, but there are obviously a lot of other uses there.We think there’s a market that’s coming at us from everywhere for the Windows CE, and you may decide in your business that you want to participate in that.And together we need to continue our feature innovations, as we’ve been doing.

Home networking is a key area, as well as the entertainment potentials that together we need to work on.And, frankly, we need to extend the user interface if we’re going to get the kind of power that we need to gain new users.

Now, switching gears for just a minute, I want to give you a quick update on where we are in terms of our product plan.Windows 98 is very close to shipping.Our plan is to ship it next quarter.In fact, assuming — assuming — nothing happens in terms of any legal issues, we will ship in the second quarter, and we will ship in June.We don’t anticipate any issues there.

Windows NT 5 is coming along fairly well — (inaudible) — 95 or 98 migration capabilities and towards a new set of distributed applications.This time around, we’re not going to say that NT is the desktop of choice for the home user, but it is the desktop of choice for business users, whether it’s laptop or not.

Now, Windows 98 — we’re pretty pleased with what we’ve been able to accomplish there.It is much more reliable than Windows 95.It is faster — it has a speed-up Wizard, which I’m sure you will appreciate.We did try to tie together the best of the PC, the TV, the games, along with the Web, and each release that we’ve done, we’ve tried to align the API, align the user interface (cluster?).So Windows 98/NT 5 and things like device driver capability have more lines than we’ve ever done before.

In terms of an actual time line, I mentioned that Windows 98, second quarter of this year.Windows NT 5 is on a hard target.It is to produce the second beta and beta II in the second quarter of this year, and we will determine when we ship given the feedback.We have our own internal target date, but it’s essentially going to be based on feedback.We don’t expect another beta after this — (inaudible) — complaints and the CD that you get here at the conference is not beta II yet.We’ve got a little bit more work to do, but you’ll see the facility is pretty good in it and a lot more features, like the intelli-mirroring — (inaudible) — are all in it.We just have a lot of other — (inaudible) — to fix.But when it is beta II, we’ll be — (inaudible) — complaints and taking feedback for — (inaudible) — to fix it the rest of the way, and then the customers will tell us when it’s ready to ship.

The key part of this line is that, as you noticed in the consumer line, we will be moving to NT technology for the consumer line after Windows 98.The way to think about it is the next version of the Windows consumer is based on Windows NT technology.

In regards to the server, we have shipped a lot of products in the past.Coming up will be — (inaudible) — the enterprise edition, as well as the server edition went into fives ships, so we’ll produce those at exactly the same time.We’ll also have updates of the Back Office and the small business server all based on Windows NT5 technology, shortly after we ship the base server in the enterprise system.

Some of the areas that we’re investing in for the future include 64 bits, Windows NT, advancing multimedia, working on natural UI — find that I mean listening — (inaudible) — as well as broad communications, making communications first, high speed; second, native to the system — (audio break) — continued hardware technology investment, which is what the rest of this conference is about.

A little bit about 64-bit NT.We do see it as a key foundation for the next century.It is part of the Microsoft scalability initiative.What we’re trying to do there is leverage all the computing power that’s available, as well as much larger memory..

It is a pretty simple move for IHP and — (inaudible) — It uses the same Windows programming models.We try to take it and just make a natural extension of it so it isn’t radically different.The major tool for all the DDIs and ADIs — and the key point here is it’s the single source for 32 and 64 bits — we have a team that is checking in live into the Windows NT 4 and NT 5 sources, the changes for NT, so that in one build environment, we’ll have actually both — 32 bits as well as 64 bits — and applications that can do the same things.

We also see this as simple for customers, because they can run 32-bit apps and 64-bit apps on the same machine, and they can enter interoperation between those for top process boundaries of com, if you will, outer process.There is a UI for 32-bit and 64-bit apps.Our plan is for a Windows NT 64, if you will, both on merced and on alpha.You will be getting a very preliminary 64-bit DEK and SDK here at WinTEch (ph).

I should be clear what this is.It’s basically just a header file and a precompiler so that you can tell through errors coming out of any of your codes whether or not you’re ready to make the change.

I encourage you to make those changes now.If you make those changes now, then it’ll just work when you move — it’ll just work once you’ve moved into the 64-bit space.

Our shipping time frame for this — it will be available on the merced time frame with full IE 64 disk support.This isn’t partial.We’ve gone through the whole system.I should say that we have gotten it up to a point where, on a simulator, we have got it to a command line (font ?), so we have to — (inaudible) —

Now, the second area I was going to mention is a technology area as well.It’s code-named CHROME.It’s an enhancement to the Windows platform.It allows new high-performance content that can come off a DVD, CD-ROM, or the Web.It exploits some of the new capabilities that we’ve got in the Windows multimedia layer, as well as new high-performance Pentium II machines.It’s a simple way to create high-end multimedia content without having to do all the programming at the direct level.You can do declarative statements and take advantage through XML of all that great power, so the Web becomes much more applicable to you.

Bob Hettle conducts demonstration od KROM.

MR. ALLCHIN: Great.Thanks.

I think the key here is this shows that if you have a rich client, the things that you can do, so instead of the environment where your client — you really couldn’t participate in the Web and the advances that are happening there.It was just too complicated.That’s pretty much changed now.You really can get all the advantages and have a high-performance client there, and that propels the industry to move faster, in our opinion.

So in summary, we hope that you really build optimized Windows systems.Obviously, Windows 98 is about to ship.We into final release, candidate’s phase, right now.We want you to take full advantage of that.We think that it will improve the customer experience because of the quality improvements there.Write our Windows NT5 drivers.We need to do that now.Be ready to sync those with the release that comes out.

We appreciate your working with us to improve the customer experience, lower that overall platform TCO to push up into the enterprise, and let’s together enable new usage scenarios to push the PC into places that it’s never been.

Of course, give us your feedback.This is an interactive conference.There’s plenty of people here from Microsoft as well as talking to your counterparts in the industry.Let’s actually drive this industry ahead.

Thank you very much.(Applause.)