Microsoft's changing tune on Linux

When Linux first appeared on Microsoft's radar a few years ago, senior company executives regularly disparaged the upstart operating system as everything from being a "cancer" to "Pac-Man-like."

And that was when they were in a kind mood.

But Microsoft's rhetorical offensive boomeranged, drawing even more attention to the open-source movement--exemplified by the Linux operating system and the General Public License (GPL) that governs it.

In the beginning, Linux was implemented for relatively low-end corporate tasks such as dishing up Web pages or storing files. But that Trojan horse entry led to unexpectedly rapid adoption elsewhere within so-called enterprise accounts. Indeed, Goldman Sachs now projects that Linux running on Intel computers is likely to become the dominant operating system in the high-end field of corporate data centers.

As LinuxWorld gets under way this week in New York, Microsoft is adopting a different tack. Gone are the over-the-top denunciations and the public broadsides. Company executives instead plan to concentrate on the technologies and to talk about the value comparisons that play up the opposing approaches represented by the Linux and Windows operating systems.

CNET News.com caught up with Peter Houston, one of the directors charged with leading the new strategy, shortly before he got on a plane to attend the opening of LinuxWorld.

Q: Why is Microsoft still struggling to come up with an answer to Linux?
A: Our thinking about how to talk to customers about Linux has clearly evolved. Two years ago, there was a lot of emotion in our discussion, and we focused on things like the GPL. We've moved it to more of a technology discussion where we want to talk about the relative merits of technology.

In practice, what will that mean?
We're talking to customers about the value of our offerings instead of talking about the shortfalls of Linux.

I would say that it's Microsoft with Windows versus IBM with Linux.

How much more of a factor in the server market do you think Linux can become?
Linux is not impacting the Windows installed base nearly as much as it's impacting the Unix installed base. There's the Windows paradigm of a comprehensive, integrated, easy-to-use stack of technologies and then there is the Unix approach, which is a piece-parts approach where the customer integrates those parts into the ultimate solution.

But that's not true overseas, where governments increasingly view Linux as an alternative if for none other than nationalistic reasons. No?
We have to separate governments as customers from governments as policy-makers. Governments as customers are as focused as anybody is on business value and price performance. The other issue is whether Linux can be a vehicle for the development of a local software economy. I don't think it's really been shown that that really is the way to grow the local software economy, though people are talking about it.

I don't hear it that way. I hear that it's more about preventing the leakage of your country's money to somewhere else--that it's more a case of how to enforce local control.
We don't compete in local economies for services business, so those dollars stay in that economy. We did a study and found that, on average, for every dollar that goes to Microsoft, nine dollars go to the local economy. And we think that's a very impressive ratio.

Still, it seems that Linux has incredible vitality and momentum--not to mention the support of IBM, Intel, Dell Computer and Hewlett-Packard, among others.
We have to look one level deeper. The biggest catalyst to Linux's growth is IBM's decision to get behind it. That has lent credibility to Linux as an enterprise offering. Why did they do it? My view is that IBM's trying to change the rules in the Unix marketplace. By commoditizing hardware based on Intel with an operating system--the kernel's very much a commodity through the open-source process--they are driving revenue to middleware and services. Look at whom that competes against on the Unix side: companies like Sun, with a virtual dependence on a hardware model and a tightly coupled OS.

No question, IBM doesn't control Linux, but it has a lot more control over it than it does over Windows. And that gives the company a lot more control over its future.
In my mind, IBM is ultimately trying to consolidate the Unix marketplace around Intel-based Linux offerings from IBM. They see that as a big opportunity though IBM's revenues from Windows out of its global services group are still significantly higher than from Linux.

Which of the advantages claimed by Microsoft do you think can't get reproduced by the Linux world?
I don't see the Linux community development model attempting to build the kind of integrated offering we have today...I think IBM made a bet that integrated software solutions will not come out of the community-based development model. They're walking that line where they are making sure they're big enough endorsers of open source so that you accept the operating system. But in fact, they are betting that the enterprise software class software you buy to deploy your infrastructure will still come from IBM as a for-charge product.

But why can't there one day be a comparable integrated operating system from IBM that's based on Linux?
That will be a long time in coming. To do it on Linux in the open-source model means they have to give up that investment in intellectual property to the community and that makes it very difficult to have a revenue-generating business.

Why does IBM have to give up its investment in intellectual property?
If they develop comparable things to WebSphere in open source-- I'm talking about just using WebSphere. I'm not talking about just an open-source version of WebSphere--but one that's integrated with Linux.
We haven't seen IBM building a level of integration between its pieces to the degree we do.

From a customer perspective, does it really matter? I think so--and that's the essence of the question. We believe that by having these pieces tightly integrated with the operating system that we can deliver benefits like single sign-on, a common management infrastructure, and consistent user-experience. All those things come from starting with an integrated operating system at the base, and then building a tools strategy and database strategy and application strategy framework strategy on top of that.

I don't see the Linux community development model attempting to build the kind of integrated offering we have today.

But in an ideal Microsoft world, there isn't much choice. You have the integrated stack from Windows to development tools, to presentation stuff, to backend SQL server stuff, etc.
If a customer builds a line business application on Windows, that will integrate into a Windows platform environment. Whether you build or buy, you still get this collection of benefits. I don't see any other operating system delivering that.

And you don't see that ever changing?
This is the crux of the argument. Microsoft is a software company. We get our revenues from solving problems in software. IBM is increasingly a services company. But it's a very difficult task to be good at making money on services and building value into your software.

So this is not so much Microsoft versus Linux as a battle between Microsoft and IBM?
I would say that it's Microsoft with Windows versus IBM with Linux. Those two have to be together. It will be about value out-of-the-box versus value created through services. It's about a rich third-partner community versus IBM wanting to have a hand itself in the ecosystem.
When I was a developer, I found that about 10 percent of the work you did was writing new functionality. The rest of the work to build enterprise-ready technology involved things like the testing and the upgrade path. The 10 percent is the fun stuff; the 90 percent is the heavy lifting. We have done a great job of learning what it means to do the heavy lifting, but I don't think the (Linux) community has focused as much in focusing on that 90 percent

Nobody likes drudgework. Agreed.
Look where IBM invests 250 programmers in engineering the kernel. A lot of that effort goes to add engineering rigor. They recognize that for Linux to be accepted as a kernel in the enterprise, it better have these attributes. But they are investing in a very small part of the Linux stack. They are putting the rest of their investment in things like WebSphere.

What's the potential risk for Microsoft in all of this?
The challenge will come if customers start rejecting the proposition that there is value in integration. We've placed a big bet on an integrated offering. We've grown our share year after year and have reason to believe this is the right approach. If it comes down to those two alternatives, we're very happy to compete for customers' business.

Two years ago, you could offer a good intellectual argument that there was a qualitative difference between Windows and Linux. Now that you're solely making it an argument about technology, it seems you have a tougher problem. A lot of people can fix technology--IBM and the rest can keep throwing billions of dollars at it until they get it right.
I still believe Linux is an extension of the Unix paradigm. It's a command-line-focused approach that's not particularly designed to be user friendly. The Windows approach is very different. I will say that the adoption of Linux is likely to be bounded by how many companies are happy with Unix. Will it have an ability to be persuasive to people that it's a more cost-effective version of Unix? Yes. For us the key challenge in 2003 will be speaking to Unix users about why they ought to use Windows on Intel rather than Linux on Intel.