Blazing the trail to Longhorn Server

Bob Muglia may have one of the hottest seats in all of Microsoft.

The desktop version of Longhorn, Microsoft's next release of Windows, may receive most of the ink. But it is the server version that is more vital to the software giant's long-run ambitions. That's because some of the most heavily touted features of Longhorn--such as mainframe-caliber computing, better security and management and networkwide search--rely on Longhorn Server.

Muglia, a 16-year veteran of Microsoft, is tasked with building Longhorn Server, likely the most complex operating system ever designed. What's more, Muglia must keep a long train of updates and service packs for older versions of Windows rolling off the production line.

Even though Longhorn Server likely won't make a debut until 2007, it's already generated controversy. Microsoft had barely acknowledged the product's existence when the company admitted that one of the software's most anticipated features, wide-scale search of corporate networks, won't make it into the first release.

Microsoft will also need to convince big companies that Longhorn's more advanced features are worth the trouble. And while Microsoft will spend the next three years building Longhorn, Linux continues to gain in popularity. Muglia sat down with CNET News.com to talk about Longhorn, the evolving Linux threat and how Microsoft builds Windows.

Q: What's Microsoft's latest thinking on Linux? The market has changed a bit in the past few years, with some consolidation. But companies continue to install Linux on servers. How does Microsoft approach that problem?
The world has changed a bit. If you went back 18 to 24 months ago, it was unclear what Linux would look like and how it would evolve. It was thought of as free. And there was a whole series of attributes that were attributed to Linux that in retrospect were inaccurate. As time has gone on, it's apparent that Linux is becoming a set of offerings from commercial vendors. When I think of Linux, I don't think about it as our competitor. I think about Linux as a technology that is used by our competitors to build competitive offerings.

There's no question about who our biggest competitor is. It's IBM.

Sometimes, those products are solutions or pieces of solutions that need to be integrated together. One of the differentiations that Microsoft has with Linux is that we are a software company first and foremost, and we think about software-based solutions to information technology problems and how our software can drive down cost. That's pretty distinct from, say, an IBM that is first and foremost a consulting company. Our focus is how to provide more out-of-the-box solutions that don't require those consulting services.

It's interesting how you characterize Linux. Bill Gates and Steve Ballmer have told us that they focus on Linux, the operating system. They frame the debate in different terms than you.
I do think that I am bringing this thinking into the organization, because it provides clarity to me. The problem with saying that Linux is my competitor is: What does that mean? What distribution of Linux (are you referring to)? If I look at a given Red Hat or SuSE Linux distribution or an IBM solution, I can look at how that's a clear competitor and how that compares and contrasts with Windows.

Looking at the world that way, who are your biggest competitors? Red Hat is becoming more aggressive. Novell has SuSE.
There's no question about who our biggest competitor is. It's IBM. Up and down, IBM is our single biggest competitor. Red Hat and Novell are competitors, because they have become the primary providers of Linux distributions. Lots of companies that are building and utilizing Linux products are our competitors. Hewlett-Packard is a great partner, and we love partnering with it when it sells Windows. And we compete with it when it sells Linux-based solutions.

Three years ago, Bill Gates told us that Sun Microsystems was your biggest competitor. How has that changed, given the agreement between you and Sun? Sun still pushes Linux pretty heavily.

There were a lot of dreams that people had inside of Microsoft for what Longhorn Server would do.

Our arrangement with Sun is still very much evolving. We are talking about ways we can work together to benefit customers. In as much as Sun is doing Linux-based solutions, that's a different kind of competitor to us as well. Sun is in an interesting situation. All of the proprietary Unix vendors are in an interesting situation, as x86 becomes more ubiquitous. So Sun is in the middle of a huge transition.

Given your focus on finding new customers with future Windows Server releases, what areas are you targeting?
In the short run, the one that we are driving the most is the small-business space. We think that there is a huge opportunity there. I also look at the midsize-business segment. It has IT generalists. We are looking at how we can do things to help those people. And we think that we can continue to grow in the networking and database areas. Also, we are putting a lot of resources into areas where people want to migrate off of Unix.

Do those areas change with Longhorn?
The list stays pretty much the same. It will change over time, because nothing is static. I think we will do a better job with branch office use and high-performance computing, but basically, it's the same list.

Will a Longhorn client come out in 2006?
People are still figuring out exact timing for Longhorn client. It goes to beta in 2005. It's six to 12 months before Longhorn Server, so maybe that's 2005 or 2006. The client guys have not given an exact date. But we're being pretty clear that Longhorn Server is coming in the 2007 time frame.

Microsoft has had trouble getting some customers to move from older versions of Windows, like Windows NT 4.0, to Windows 2000 and Windows Server 2003, the latest server version. Where does that stand now? In the last 12 months, about 35 percent of the base has moved to Windows 2000. It's accelerating. We will see in this calendar year another third of the base move. It's a pretty small percentage of customers on NT 4.0--less than 20 percent. Japan is higher than that. The United States is lower. But the vast majority of customers will move by the end of this year.

What percentage will take advantage of 64-bit versions of Windows Server?
A very high percentage. It depends on how fast the hardware ships. Any application with a high memory demand will see the advantage of 64-bit.

What about WinFS?
WinFS is in Longhorn server. Some of the functionality of WinFS and some of the scenarios of may be limited in terms of what it can do. We will use WinFS, clearly, for the shell in that environment. WinFS will be there for collaborative scenarios. But I don't know that we will have all of the scale to the level where we would like to have it, so that you could use it for very high-volume enterprise servers. So it could be that, for example, you could use WinFS as a server for collaboration in work groups. But if you want to support hundreds of users, that may wait for the update release. One of the great things about doing these releases on a regular basis is that every two years, there is an opportunity to improve things.

Jim Allchin told us that some features of Longhorn won't be in the first release. Was he referring to WinFS?
There were two aspects that Jim was talking about. One is areas of maturity, associated with what you would expect from an enterprise-class file system that we are going to continue to work on. The other is that there were a lot of dreams that people had inside of Microsoft for what Longhorn Server would do. There is a natural process, whereby as a release transitions from the early dream stage into the reality stage, in which functionality and the scenarios get cut back. Jim was referring to some of those things. That's part of the natural process that every release goes through.

Will those features be in the next major release of Windows Server, code-named Blackcomb? Or will they be in an update release to Longhorn?
I expect that the update release of Longhorn Server will be the time when we will be able to mature much of WinFS.

How much of Microsoft's Windows development team is working on, say, Longhorn as opposed to the R2 update release to Windows Server 2003?
It's hard to answer exactly how much of our resources are devoted to R2 versus Longhorn. Our branch office team, for instance, is almost 100 percent focused on R2. The interface development team is almost 100 percent on Longhorn. The release teams are working on things in parallel, but have a tendency to focus on short-term things. Virtually all of the release resources are on XP SP2 (Service Pack 2) now, which will ship this summer, and then on SP1 (Service Pack 1) (for Windows Server 2003), which will ship later this year. If you had to pin me down for a number, I'd say it's 50-50 when all is said and done. But it varies so much by teams.

Have you begun to think about Blackcomb, even though it likely won't debut until next decade?
In fact, I'm starting to think about Blackcomb. I've been meeting with teams, saying, "Let's talk." If Indigo (a major feature of Longhorn) took four years to develop, some major infrastructure things inside Blackcomb will also take four years to develop. The amazing thing about that is that there isn't much time to get that done...It takes an airplane or a space subsystem or something to compare with developing Windows, in terms of complexity associated with some of these things. We have this huge set of interoperability and other things that we need to think about, as we move things forward.

Can you tell us what's on Blackcomb's feature list?
My top priority is continuing to drive the Dynamic Systems Initiative (for systems management) part of the equation. While we have all of these great short-term deliverables for DSI, there is a lot more we can do. We're taking the concept of transferring information across the life cycle of the business application and ingraining it in as part of the process. DSI is all about information transfer between a developer, the operations center and the end user. There are ways to do that on a surface level, and there are ways to build that deeper into the OS, and that's what we are doing.

Bob Muglia may have one of the hottest seats in all of Microsoft.

The desktop version of Longhorn, Microsoft's next release of Windows, may receive most of the ink. But it is the server version that is more vital to the software giant's long-run ambitions. That's because some of the most heavily touted features of Longhorn--such as mainframe-caliber computing, better security and management and networkwide search--rely on Longhorn Server.

Muglia, a 16-year veteran of Microsoft, is tasked with building Longhorn Server, likely the most complex operating system ever designed. What's more, Muglia must keep a long train of updates and service packs for older versions of Windows rolling off the production line.

Even though Longhorn Server likely won't make a debut until 2007, it's already generated controversy. Microsoft had barely acknowledged the product's existence when the company admitted that one of the software's most anticipated features, wide-scale search of corporate networks, won't make it into the first release.

Microsoft will also need to convince big companies that Longhorn's more advanced features are worth the trouble. And while Microsoft will spend the next three years building Longhorn, Linux continues to gain in popularity. Muglia sat down with CNET News.com to talk about Longhorn, the evolving Linux threat and how Microsoft builds Windows.

Q: What's Microsoft's latest thinking on Linux? The market has changed a bit in the past few years, with some consolidation. But companies continue to install Linux on servers. How does Microsoft approach that problem?
The world has changed a bit. If you went back 18 to 24 months ago, it was unclear what Linux would look like and how it would evolve. It was thought of as free. And there was a whole series of attributes that were attributed to Linux that in retrospect were inaccurate. As time has gone on, it's apparent that Linux is becoming a set of offerings from commercial vendors. When I think of Linux, I don't think about it as our competitor. I think about Linux as a technology that is used by our competitors to build competitive offerings.

There's no question about who our biggest competitor is. It's IBM.

Sometimes, those products are solutions or pieces of solutions that need to be integrated together. One of the differentiations that Microsoft has with Linux is that we are a software company first and foremost, and we think about software-based solutions to information technology problems and how our software can drive down cost. That's pretty distinct from, say, an IBM that is first and foremost a consulting company. Our focus is how to provide more out-of-the-box solutions that don't require those consulting services.

It's interesting how you characterize Linux. Bill Gates and Steve Ballmer have told us that they focus on Linux, the operating system. They frame the debate in different terms than you.
I do think that I am bringing this thinking into the organization, because it provides clarity to me. The problem with saying that Linux is my competitor is: What does that mean? What distribution of Linux (are you referring to)? If I look at a given Red Hat or SuSE Linux distribution or an IBM solution, I can look at how that's a clear competitor and how that compares and contrasts with Windows.

Looking at the world that way, who are your biggest competitors? Red Hat is becoming more aggressive. Novell has SuSE.
There's no question about who our biggest competitor is. It's IBM. Up and down, IBM is our single biggest competitor. Red Hat and Novell are competitors, because they have become the primary providers of Linux distributions. Lots of companies that are building and utilizing Linux products are our competitors. Hewlett-Packard is a great partner, and we love partnering with it when it sells Windows. And we compete with it when it sells Linux-based solutions.

Three years ago, Bill Gates told us that Sun Microsystems was your biggest competitor. How has that changed, given the agreement between you and Sun? Sun still pushes Linux pretty heavily.

There were a lot of dreams that people had inside of Microsoft for what Longhorn Server would do.

Our arrangement with Sun is still very much evolving. We are talking about ways we can work together to benefit customers. In as much as Sun is doing Linux-based solutions, that's a different kind of competitor to us as well. Sun is in an interesting situation. All of the proprietary Unix vendors are in an interesting situation, as x86 becomes more ubiquitous. So Sun is in the middle of a huge transition.

Given your focus on finding new customers with future Windows Server releases, what areas are you targeting?
In the short run, the one that we are driving the most is the small-business space. We think that there is a huge opportunity there. I also look at the midsize-business segment. It has IT generalists. We are looking at how we can do things to help those people. And we think that we can continue to grow in the networking and database areas. Also, we are putting a lot of resources into areas where people want to migrate off of Unix.

Do those areas change with Longhorn?
The list stays pretty much the same. It will change over time, because nothing is static. I think we will do a better job with branch office use and high-performance computing, but basically, it's the same list.

Will a Longhorn client come out in 2006?
People are still figuring out exact timing for Longhorn client. It goes to beta in 2005. It's six to 12 months before Longhorn Server, so maybe that's 2005 or 2006. The client guys have not given an exact date. But we're being pretty clear that Longhorn Server is coming in the 2007 time frame.

Microsoft has had trouble getting some customers to move from older versions of Windows, like Windows NT 4.0, to Windows 2000 and Windows Server 2003, the latest server version. Where does that stand now? In the last 12 months, about 35 percent of the base has moved to Windows 2000. It's accelerating. We will see in this calendar year another third of the base move. It's a pretty small percentage of customers on NT 4.0--less than 20 percent. Japan is higher than that. The United States is lower. But the vast majority of customers will move by the end of this year.

What percentage will take advantage of 64-bit versions of Windows Server?
A very high percentage. It depends on how fast the hardware ships. Any application with a high memory demand will see the advantage of 64-bit.

What about WinFS?
WinFS is in Longhorn server. Some of the functionality of WinFS and some of the scenarios of may be limited in terms of what it can do. We will use WinFS, clearly, for the shell in that environment. WinFS will be there for collaborative scenarios. But I don't know that we will have all of the scale to the level where we would like to have it, so that you could use it for very high-volume enterprise servers. So it could be that, for example, you could use WinFS as a server for collaboration in work groups. But if you want to support hundreds of users, that may wait for the update release. One of the great things about doing these releases on a regular basis is that every two years, there is an opportunity to improve things.

Jim Allchin told us that some features of Longhorn won't be in the first release. Was he referring to WinFS?
There were two aspects that Jim was talking about. One is areas of maturity, associated with what you would expect from an enterprise-class file system that we are going to continue to work on. The other is that there were a lot of dreams that people had inside of Microsoft for what Longhorn Server would do. There is a natural process, whereby as a release transitions from the early dream stage into the reality stage, in which functionality and the scenarios get cut back. Jim was referring to some of those things. That's part of the natural process that every release goes through.

Will those features be in the next major release of Windows Server, code-named Blackcomb? Or will they be in an update release to Longhorn?
I expect that the update release of Longhorn Server will be the time when we will be able to mature much of WinFS.

How much of Microsoft's Windows development team is working on, say, Longhorn as opposed to the R2 update release to Windows Server 2003?
It's hard to answer exactly how much of our resources are devoted to R2 versus Longhorn. Our branch office team, for instance, is almost 100 percent focused on R2. The interface development team is almost 100 percent on Longhorn. The release teams are working on things in parallel, but have a tendency to focus on short-term things. Virtually all of the release resources are on XP SP2 (Service Pack 2) now, which will ship this summer, and then on SP1 (Service Pack 1) (for Windows Server 2003), which will ship later this year. If you had to pin me down for a number, I'd say it's 50-50 when all is said and done. But it varies so much by teams.

Have you begun to think about Blackcomb, even though it likely won't debut until next decade?
In fact, I'm starting to think about Blackcomb. I've been meeting with teams, saying, "Let's talk." If Indigo (a major feature of Longhorn) took four years to develop, some major infrastructure things inside Blackcomb will also take four years to develop. The amazing thing about that is that there isn't much time to get that done...It takes an airplane or a space subsystem or something to compare with developing Windows, in terms of complexity associated with some of these things. We have this huge set of interoperability and other things that we need to think about, as we move things forward.

Can you tell us what's on Blackcomb's feature list?
My top priority is continuing to drive the Dynamic Systems Initiative (for systems management) part of the equation. While we have all of these great short-term deliverables for DSI, there is a lot more we can do. We're taking the concept of transferring information across the life cycle of the business application and ingraining it in as part of the process. DSI is all about information transfer between a developer, the operations center and the end user. There are ways to do that on a surface level, and there are ways to build that deeper into the OS, and that's what we are doing.