What does the platform strategy group do?

We spend time trying to educate customers in the marketplace when we think we've got a pretty good story to tell.

Then I've got a technical team. I've hired quite a few people from the Linux open source world to run a research and development centre.

Then we bring consultants from the Linux open source space to help us build solutions. We've got every Linux distro running; about 120 servers all with varying degrees of open source stacks on them.

What we do there is help our product groups have line-of-sight when they need to know things about Linux.

Our developers have been heads down on Microsoft for years and they don't know some things that might be attractive or appealing in the Linux space. So we try to give our developers line-of-sight to help our product road maps.

We also add credibility to our externally-facing activities. For instance, when I say our file serving can outperform Samba, [it] is not a gut reaction but because I've had Linux and open source guys build the software environment [and] my Windows guys build the Windows environment.

We've realised that we need more direct interaction with customers and in our individual markets. Because, just as total cost of ownership is very unique based on customer scenarios, there are cultural and climate differences in EMEA versus other areas.

So we have a small team of people around the world charged with the same initiatives I have, but in a more localised way.

How about Longhorn? How is that going at the moment?

We made the announcement in August to take WinFS out of the 2006 release, and have WinFS available shortly after that.

It caused a bit of a buzz obviously, but on this notion of Linux and open source compared to a platform, people say: 'Don't you need Longhorn to compete with Linux?' And I say: 'No, nothing could be further from the truth. From an innovation perspective, a technology perspective, I've got everything I need today from a value platform.'

So why do you need to do so much work?

Longhorn is a huge bet for the company on a variety of fronts. But it has nothing to do with Linux and open source and everything to do with us continuing to innovate.

We're re-architecting the messaging bus, the interface, the storage system. We're re re-architecting many things. We're also providing avenues and paths for other products to take advantage of the operating system, such as WinFS, or Indigo or Avalon or whatever the case might be. That's a big undertaking.

Isn't there a danger of having an enormous ever-growing monolith? Shouldn't it be much more modular?

We look at it less [as] modularity, and more as role-based deployment. For instance, whenever you have a loosely coupled environment, end users have more cost trying to tighten that up, to run their solutions [and] operations, to manage that, to patch that, all those types of things.

You can take a loosely coupled environment and say: 'We're just going to deliver a set of bits that are modular by nature, then the end user or service provider or whatever can tie those things together as a solution'.

We say we can deliver a tightly coupled environment to market. We can support that better [and] people can build applications that can run faster.

[It's] not so much that people just want a piece that's modular, but more like: 'What are you trying to do with the technology?'

What we found from the server side, especially in the enterprise, is that people want to dis-aggregate their workloads. So yes, we ship a multi-purpose operating system but we know they want to deploy maybe just a web server, or file server, or high performance cluster environment.

So one of our Longhorn design points is to allow role-based deployments. So you basically provision and deploy what you want without having the trade-off of manageability, security and the other concerns you have [with] a more loosely coupled modular environment.

How are you seeing Linux in the market?

The other thing we're finding is more and more people wanting to deploy a commercialised Linux version. They don't want their own custom configured kernel [or] custom distribution. They want to pick up the phone [and say]: 'Help me this is broken.'

So that puts you into Red Hat/Novell-SuSE's ballpark. Both those have pricing models for support and security patches more expensive than Windows Server. So in some cases you could say I am under-priced compared to the marketplace. That's what we're seeing.

What about applications such as Great Plains and Navision sitting on top? What is your strategy moving up the stack?

The broad strategy with Navision and Great Plains is more about the engine. We think about the entire IT ecosystem, and we see a set of repeatable processes that a lot of ISVs are having to do, in this case in some of the traditional accounting systems in mid-tier and small-tier type scenarios.

And so we [ask] what if we could actually build an engine leveraging the assets of Great Plains and Navision to allow ISVs to build on top of to reduce their research and development costs and hopefully increase their time to market with applications, and also have a better integration story within Windows?

The long-term strategy is to continue to hone and facilitate the ecosystem from that perspective. That was the central design point in both those acquisitions.

But since you're getting into what ISVs do, how happy are resellers about that?

Navision resellers are happy because they're now part of the broader Microsoft reseller community. So now they have the opportunity to do things they didn't have before.

The same with Great Plains resellers. I spent a week in Fargo with a bunch of Great Plains resellers having this conversation a few years ago. They're getting access to a lot of [new] customers. They're excited about that.