IBM's Steve Mills Talks Software Strategy

Steve Mills, senior vice president and group executive of IBM Software Group, sat down this week with CRN Industry Editor Barbara Darrow and Editor in Chief Michael Vizard to discuss IBM's middleware strategy, SMB game plan and the competitive landscape.

CRN: Microsoft last year made a big deal about an upcoming integrated Web application suite and now scrubbed it saying people don't want an integrated suite. It was viewed as an anti-WebSphere move. Why do you think they scrubbed it?
Mills:: You can never tell for sure what's going on there, anymore than they can tell what's going on inside IBM. I think that the issue of middleware at Microsoft in general has always been a controversial topic. They always see the alternative model, characterized by open systems and what evolved over many years in the Unix space to a modular system structure, someone provides the hardware and the operating system, the database comes in separate and other mechanisms. It's a different type of approach to creating a systems architecture and one with well-insulated layers and the ability to snap things in and out. It provides portability, and I think in general they don't like that model [laughs]. Because it obviously creates opportunities for others to come in and disintermediate your control of the stack, as it were.

Microsoft would prefer the operating system to be the anchor point and everything [else] built onto that anchor point. SQL Server does not exist without Windows. It needs native Windows services. The original code base coming in from Sybase was optimized deeply down into Windows. Elements were eliminated. The Windows task management, memory management, other control services are now an interdependent element with SQL Server. And the same thing applies to MSMQ, MTS and other mechanisms that were part of the BackOffice stack, some of which they don't talk about much anymore. [They like] that whole model ... from a business perspective. On the flip side, it means huge quantities of code have to move forward in lockstep, which makes it very challenging to get new things out. We waited forever for Cairo and it never arrived.

I think this is an internal conflict in Microsoft, and I think that the operating system-centric approach seems to win out.

CRN: What did you think of the Microsoft-Sun deal, given all that?
Mills:: I think it was a legal settlement. It's a $2 billion payment, and obviously everyone knows Microsoft has many, many billions of dollars. In the scheme of things and given the range of legal challenges they face, I assume they felt it was a good use of money.CRN: You don't see it as focused against [IBM]?
Mills:: No. More has been written and speculated about on this settlement than either company has said. They have not been very specific on what collaboration could unfold. It's pretty clear it's a legal settlement. Microsoft is not interested in promoting Solaris in the market. Sun is very much a hardware company.

CRN: Don't they both have mutual self-interest in minimizing the value proposition of the app server by collaborating together on integration work, incentive to work together to minimize BEA Systems?

Mills:: Perhaps. But to me that's fighting mother nature. It doesn't really make a lot of sense. The term app server is a 1990s euphemism for a transaction monitor manager. The concept of a transaction monitor manager has been with us since the 1960s and proven valid an endless number of times. It's the way you get scalability. It's the way you get the ability to manage lots of applications concurrently. There's intrinsic value there. It's a core computer-science construct and so it's required capability. How you package and deliver can be in a variety of ways. You can obviously tie anything to anything from a packaging perspective ... and it's a great mechanism for incorporating integration technology because it is, in some respects, almost a network operating system-type mechanism. It's designed to deal with traffic flow, traffic movement, and if you follow the right standards, it talks to itself very well, passes traffic back and forth. It's a network operating system, and if you want to deliver networked systems, which are becoming increasingly popular through Web-based scenarios around network commerce, connected commerce, multitiered systems needed to deliver commerce applications coming through the second half of the '90s. Certainly these more elaborate context-passing of transaction integration solutions lend themselves to the use of this type of mechanism. I don't know how you get around not having one. You have to build one. Sun's not been very successful in their attempts through acquisitions of NetDynamics and subsequently by Kiva via the AOL-Netscape deal. They haven't made much of a market here, and yet there is clearly a big market here. IBM and BEA Systems do very well at this.

CRN: Couldn't they build that functionality in common with each other at the top of their stacks?
Mills:: Well, it's a lot of work to build this stuff. It has to do all the things I said. You could decide to deliver one for free I suppose, but free doesn't help if it doesn't do the job.

I don't see either company focused on creating this type of mechanism and investing the years of developers. Can you imagine Sun and Microsoft developers working together for years?

CRN: How do you manage the challenge of going into vertical-market segments? It makes sense at first, but if you peel it back, you have maybe five major product groups, and each of the five has 20 products, and now each will get vertical spins. It's very complex.
Mills:: The divisions are not separate entities. It's one architecture, one process, one management system. Nothing's surfaced in this organization without coordination from above. We're not just delivering things for the fun of it. All of these solutions are composites of technology from each of the divisions. It's kind of like baking a cake. The divisions do ingredients and solutions are a composite of the ingredients. You've got to deliver collaboration, Ambuj [Goyal, general manager of IBM Software] has the collaboration dimension. You need transaction control, [General Manager of IBM's WebSphere] John Swainson has that. You need data management/integration, [General Manager of Data Management] Janet [Perna] has that. [General Manager of IBM Tivoli Software] Robert [LeBlanc] has systems management and security. And [General Manager of IBM Rational] Mike [Devlin] has tooling. Each plays their role.There are certainly buyers out there who buy in pieces, so products remain a valid paradigm of putting things into the marketplace but the shift toward more of an integrated package with more preintegration, more certainty of success in deployment, is building in popularity. We're just baking cakes.

CRN: How happy are you with the Express line and take up? At one point, you said something like 10 percent to 15 percent of the base is WebSphere Express, but a lot of people look at that and then buy full-boat WebSphere. What is the goal for Express? Is it really necessary or do you want to move it downstream?
Mills:: Ten [percent] to 15 percent still remains a valid statement of where we are. I'm trying to engage with a wider collection of customers and partners, trying to build out the ecosystem. That is a core goal. As we get into midsize customers, at the smaller end of the midsize market, the skills begins to diminish pretty rapidly. The dependency of application vendors and local business partners to deliver business capability goes up dramatically. We think Express packaging gives us leverage for that class of customer, for the partner at the bottom end of the market for us. If the ISV or the business partner gets attracted to working with IBM because of Express and decides to flip to the full version, fine with me. I'm doing Express because I think I need Express to garner attention at that end of the market, to be able to demonstrate easily and quickly that this technology is not that hard to work with. Like anything else, it seems daunting at first if you have no familiarity with it. How can I provide a basis for someone to get into it, have a little success with it, see that we're truly dedicated to that end of the market? And then if they go with the full version to meet new requirements, so be it.

CRN: How can you juice Express for the lower end of the market. Is it a marketing thing or a product thing?
Mills:: We've shifted a large percentage of our marketing budget over to midmarket messages and Express. IBM across the board is adopting this Express term for midmarket, so what began in the software group is picking up elsewhere because they see the effects of this. You have to qualify and package to be Express. The secret to success at that end of the market is the application vendors, the ISVs. There's nothing more important than ISVs there. It's an entirely application-led sale, so we need more midmarket ISVs, and we've been spending lots of money, scouring the landscape, porting and enabling. It becomes very localized. We've been doubling and redoubling our business partner and territory management sales teams whose job is very much geared to how they manage their partners to bring capabilities to customers. If I run midmarket territory in Chicago and try to manage an ecosystem many times larger than my own resources through partner relationships, that's how I get the bulk of my sales. In order to have a partner relationship you have to deliver something with IBM or it doesn't make any sense. This is where Express packaging and ISV enabling comes into play. This is critical. This is what drives this market.

CRN: How often do you get questioned by ISVs about where your middleware ends and applications begin?

Mills:: It's really no problem whatsoever. We're not delivering any business logic function.

CRN: If you talk to Information Builders or some others, they're pretty open about feeling competition with IBM. Mills:: If you talk to those delivering applications and not middleware, they know IBM doesn't do that, not doing accounting, accounts payable, CRM, supply chain, inventory, billing systems. A long laundry list. If you talk to companies in the middleware business, information integration like Information Builders, well, guess what? I'm in the information integration space. I acquired the assets of CrossAccess to provide connection capability into a variety of data sources. Information Builders finds some of these things overlapping with what they do. We are a middleware competitor. So BMC [Software] doesn't like me, Oracle doesn't like me, [Computer Associates International] doesn't like me, BEA doesn't like me. Lots of companies don't like me. I'm not asking for them to like me, and I'm not disappointed.CRN: Do you see the ISV market going more global?

Mills:: If I look at Top 10 packaged apps, SAP etc., that's global. If I go below that to midmarket apps, many are regional, but they come together on platforms, it should become easier for them to go global. I think the barriers will drop. The standards and standard infrastructure are adopted. But I think the go-to-market capacity of these companies is more significant than the technology barriers. It can get easier perhaps from the tech perspective but these companies are spending 40 [percent] to 50 percent of their revenue on go-to-market, that tends to be the big inhibitor. Expanding into another geography might be overreaching.

CRN: Isn't that where your salespeople could help?
Mills:: But I don't want to be in the applications business. We'll participate in joint sales with a software vendor, but we don't want to be their remarketer. Then I have to train people and compensate my reps. My reps won't sell anything they're not paid on.

I don't want to be the lead seller of the app. I'm happy to have the local, regional-based. I'm happy to work with them locally. Most business is really local anyway. This global reach stuff that some guys want to play in, they're the minority.

People fail to recognize if you take out Microsoft's productivity stuff, the applications market was in the $100 billion range of end-user expenditures. SAP is biggest at $4 billion. If you add up the Top 10 biggies, 75 percent of the market is still made up of smaller companies. It's a huge small-business market for producing applications. And many of them are country-specific, many are industry-specific, they may be pan-geography but industry-narrow. Selling to money center banks, there are five-to-six money centers in the world.

CRN: You're not in the applications business, and yet every time I read about IBM and CAD/CAM, I read about IBM and Dassault. It's the poster child. If I'm Autodesk or whatever, why would I want to partner with IBM when it's blessing my competitor?

Mills:: You might not. Catia is the product. We made that decision a long, long time ago. That decision goes way back to the '80s. And that model in that particular narrow niche of the market works for us. We like it. We live with it. I don't think the Catia relationship is perceived by any ISVs as an indicator of IBM's intent to become an application vendor.CRN: Can't you figure out a way to incent IBM reps to sell third-party apps? If they generate demand?
Mills:: If an app licensed for a million dollars generates $5 million in related hardware, software and services for IBM, then the rep is highly motivated. There is so much technology and services associated with all these apps. This silly Larry Ellison thing, [where he says] 'You buy out of the box, services is a sham thing'--well, hundreds and hundreds of millions of services revenue comes to IBM from Oracle apps. There's no instance of an Oracle app that goes in out of the box. Our reps get compensated on sales of associated services and IBM product sales.

We've built this joint-sales process with many of these ISVs, once the field-sales team starts to see what the benefits of that are, they're highly motivated to want to play. It's well recognized as the way to go.