IBM generated $13 billion in enterprise software revenues last year. So, when Steve Mills, senior VP and group executive with IBM Software Group, talks, even Microsoft tends to listen.

Microsoft Watch Editor Mary Jo Foley chatted with Mills a couple of weeks after he appeared on stage with Microsoft CEO Bill Gates to demonstrate Web services interoperability. Mills had plenty to say about the Redmond software vendor.

Read About the Gates-Mills Dog and Pony Show Here

Here's our transcript, edited for length:

Microsoft Watch: To start, a Sun question for you. Sun recently announced its plan to provide an integrated suite of server software, called the Java Enterprise System. Do you think Sun is simply copying what IBM's already done around WebSphere?

Mills: I think Sun doesn't really understand what a software company does. A successful software company has to have a software sales force, support in the field for software. Software is sold at an entirely different rate and pace. It has very different sales characteristics than hardware. So this idea of sort of "packaging it together and buy it with the hardware" idea is not how the market buys. Certainly not in the enterprise space.

There are two buying habits out there that remain valid. One buying habit is the shrink-wrapped phenomenon. The other is enterprise institutional buying, which is really more about the enterprise trying to achieve some business goal through deploying some fairly complicated pieces of software; making them work together with some other fairly complicated pieces of software; and six months after the purchase, being able to demonstrate a result that means something to the business.

So, if you don't have a field organization that understands the customer issues and relates the technology to their business needs; you don't have the resources to help the customer deploy it, tune it and get it to work right, you don't have a successful software model. And I don't see a lot of evidence that they are building the go-to-market structure around software. They're sort of presuming that terms and conditions alone will carry the day. And maybe there are enough people out there that hate Microsoft — which is not much of a motivation to buy anything. Sun may hate Microsoft, but the rest of the world doesn't quite think of them that way.

Microsoft Watch: So contrast that Sun approach with what Microsoft's talking about around "integrated innovation." Microsoft is talking software stack, too. Do you see Microsoft's approach as differing from Sun's...and IBM's?

Read More About 'Integrated Innovation' Here

Mills: I see (Microsoft's approach) as a continuation of BackOffice model. You can't equate these companies together. Microsoft has been able to do things because of reasons of history. A set of unique events occurred and they were the ones who lucked out.

But the PC model is a very different model than the enterprise server model. Microsoft clearly leverages the PC model extremely well. They'd like to translate some of that to the server business…. For lower-end buyers still doing a lot of client/server computing, these integrated packages can work. We're doing some of it ourselves. It's a valid model, but we're not naive about how far it stretches.

Microsoft Watch: But do you think Microsoft is trying to do what IBM is, with its WebSphere model, with Integrated Innovation? Or not exactly?

Mills: Not exactly. Architecturally, they have a hard time with WebSphere. It's a foreign body. Not only do they not do it, they actually don't want to do it.

One of the first principles at Microsoft is the operating system is the control point....the OS is the session-control service. But WebSphere says I am going to take over supervisory control, user connection, session control, arbitration, back-end connection, database connection. The database actually is a passive entity in the WebSphere model. So they really don't like the WebSphere model. If they did, they would have let Jim Gray loose to go build a real MTS (Microsoft Transaction Server), instead of what they built.

Who is Jim Gray? (If You Need Ask)

We love the WebSphere model because we love IMS and CICS. This is the favorite IBM model because we love this layered architecture that, in the distributed world, gives us OS insulation. So we make it portable and get a lot of derivative benefit from the model in the distributed computing environment, beyond what we got out of it in the mainframe environment.

So, I don't really know where they are going there, exactly. I think they are struggling with some of this stuff architecturally. There are strong religious beliefs among the different technical leaders at Microsoft. They are not very nice to each other. I don't think Bill (Gates) understands WebSphere, transaction monitor manager, insulating layer. He doesn't connect with it mentally. And he's the only one who ever would direct the inmates to ever change their model. Because (Platform Group VP Jim) Allchin would never do it.

(It was a different world) when Unix was fragmented and standards didn't really take root.... But that's no longer a customer discussion. Now, with the Internet and all the standards that have emerged on top of it, we've come to conclude that, yeah, interoperability works.

Bill (Gates) hates Linux because Unix was supposed to go away. That was the model. I've said this to him.

Microsoft Watch: And what did he say?

Mills: He shook his head (indicating yes). You don't like it because Unix was supposed to go away. We like it because we are deep into the Unix business. We want to see that franchise continue. We don't want to give up all that business that we built there.

Microsoft Watch: Could you talk about what was behind your recent appearance with Gates in New York to demonstrate the fact that your evolving Web services are interoperable? I mean, you've been working with Microsoft for some time and have demonstrated several times before that you are working toward interoperability and coexistence. So what was new, this time, if anything?

Mills: We've been working together for three years. The motivation is both similar and different for each company....

It's a gambit. It's a play. Gates has sort of a more practical, co-opetition type mindset. (CEO Steve) Ballmer can't stand to be near IBM. So, we're not fooled by the attitudes that are there. But this could be a way to get this whole XML thing moving in the marketplace that could be very leveragable for us. It's important that we demonstrate "any client to any server" kind of capabilities. So, getting Microsoft in the game and settling out the SOAP, WSDL, interfaces, reliable messaging will be a big incentive to the marketplace to do more.

Microsoft Watch: With all of the new Web services-related technologies that Microsoft is talking about adding to Windows in the Longhorn and post-Longhorn timeframes — things like the Microsoft Business Framework, Indigo, etc. - do you think there's a danger that Microsoft might say "OK, we've done our part and now we're going our own way"?

Mills: I don't think so. First of all, I'm already doing those things today. I've got 48 different industry-branded versions of WebSphere doing all kinds of things like EDI mapping, specialized device support, we attach the public-switch telephone network, we're doing stuff with Rockwell Automation, Honeywell. We're doing HIPAA, FDA - all manner of interfaces. So we're already into all this industry-vertical stuff, big-time, from a middleware-enabling standpoint.

I expect them to do the same. They aspire to similar things that we do. The difference is we're not waiting for Web services. We're doing hard-coded C adapters, Java adapters, every kind of adapter imaginable as a way of getting the customers going today rather than waiting for some future emergence of everything being more perfectly mapped using XML.

So I'm not really concerned about what they're trying to do. It's sort of expected. But will they diverge? Well, the thing that permits divergence is a lack of market acceptance. The market is already doing the Web services infrastructure, per the roadmap that we (Microsoft and IBM) have laid out. That roadmap is anything but complete. It deals with some core, very fundamentally challenging areas. But beyond the BPEL first point of intersection with process - which is a fairly neutral sort of mechanism in terms of the semantics of BPEL - all that is extensible into vertical-industry standard spaces.

There are other untouched areas in certain aspects of establishing policies, rules — the list goes on and on. We haven't broken ground on those things, and it's not clear that our interests exactly dovetail about the level that we're at today.

Microsoft Watch: So, you may not do more together?

Mills: We may not go beyond our current roadmap. There is no additional roadmap to reveal. Let's put it that way. We'll finish up the work that we're doing and talk about other work that we could do. But we may reach an endpoint, or a point of completion, if you will, that is satisfactory to both companies, and kind of move on from there.

It's a very pragmatic effort. We haven't suddenly become Siamese twins here.

See the Current MS-IBM Web Services Roadmap (From 2001) Here

Microsoft Watch: Switching gears, has Windows become a liability for you and your customers? Are customers coming to you, saying, "I just want off Windows. I just can't deal with all these bugs and patches and viruses"? And what are you suggesting they do?

Mills: Part of the problem is that they are being attacked. No product is perfect. Their product is a target, and that's a big part of the problem. You have malicious people who spend all their time figuring out where the back doors are, and what the scenarios are that you can take advantage of - and they are exploiting them. The world's not fair. In some respects, this is a little bit unfair. But they are a company that kind of sets themselves up to be a target as well. It's not a Mister Rogers image. It's much more in your face. The Microsoft Way. That approach that they have causes them to be a target. And that's not going to change.

You are going to have customers getting angrier and angrier over this. I don't know where the break point is. It's a real slippery slope that we're on. If the attackers continue at the current rate and pace, Microsoft's going to have to deal with this at the customer level. Of course, Microsoft certainly assumes no liability. It's a tough problem.

Microsoft Watch: So, what are you advocating that your customers do?

Mills: People are angry about Windows. But we don't have an easy answer to this. In many businesses, the number of thick Windows clients that they run is more than they need to run. They could improve their operational efficiency, reduce their administrative costs by looking at which users they want to move to a thin-client model. And that would diminish the number of attack points. It wouldn't eliminate Windows as an attack point. But it would reduce the number of attack points. When you run a thin client, the OS is not a critical element. It just as easily could be Linux-Mozilla and Windows-IE (Internet Explorer) in a thin-client environment.

Microsoft Watch: So, is this an oversimplification? At the same time that Microsoft is pushing for its fattest client ever with Longhorn and Office 2003, IBM is pushing for thinner clients?

Mills: Longhorn is going to be one big client. Longhorn is way out in front of what will be the typical desktop size at the point in time in which Longhorn launches. And replacement cycles are getting longer. Businesses are saying, "do I really want to do all this work on the desktop? Is this what my employees need?"

I think they pay too much attention to the U.S. market, frankly. The U.S. market has a sense of entitlement. Of course you have to give people the latest PCs and all the software and office supplies. Outside the U.S., the attitudes are very different. I think they're missing the business-buyer issues.

Rich client is still valid. I carry a rich client around. And I'm likely to continue to do that. For many people in the IBM company, given what we pay them, we're going to give them the tools that match the productivity that we're trying to get from them. For many other people, though, what do they need it for? I think there are a lot of other possibilities for people who are clerical workers, people who are in branch banks, plant floor — very line-of-business, task-specific applications. Should I give them a thick client, or move to some kind of hybrid: some thick, some truly thin and some that may be a little bit in between?

(This article is excerpted from the October 2 and October 6, 2003, issues of the Microsoft Watch newsletter.)