Tech Update: IBM is a real big place. What's your role here?

Mills: I'm the software guy.

Tech Update: In the context of the portfolio that IBM has, that includes what?

Mills: The bulk of our investments are in middleware technology. So WebSphere, DB2, and the database technologies, Tivoli systems management, Lotus collaboration. These are IBM's largest investment areas. We also invest in operating systems. We have operating systems that are specific to our hardware platforms and they're a piece of my software portfolio. If you take the $14 billion of total software revenue we have, $11 [billion] of it is middleware and about $3 billion of it is operating systems.

Tech Update: If you were sitting in a room with the members of our audience who are faced with the question of how they're going to do business in the 21st century, what would you tell them are the five most important priorities that they should be paying attention to?

Mills: There's really one priority objective with a lot of items nested under it--the creation of an integrated infrastructure that ties together the business processes in whatever value chains make sense to the way you run your business. It's the idea of creating an integrated demand-to-delivery environment. It's straight-through processing. It's leveraging the technology to give you the best possible connection for your customers through your internal processes out to your suppliers and back again. Information technology is a tool; it's a means to an end. What you're trying to accomplish is to get to a much more efficient way to do business, thereby improving your financial ratios.

Tech Update: What's a good example of where demand-to-delivery and straight-through processing serve as means to the ends of improved efficiency and financial ratios?

Mills: Every customer contact is a moment of truth for a business. Businesses need to serve their customers quickly and effectively with the right information at that point in time. They must drive the customer's request through their internal processes and back end. That typically involves inventory, manufacturing, replenishment, and supply chain activities, which have their logical correlations in the finance and insurance industries as well. If businesses can line those things up effectively, they're going to be more efficient. Therefore, you improve your financial ratios, your bottom line looks better, and your shareholder value increases. You're just using technology as a means to an end to get there.Tech Update: Integration isn't a new idea though. It's been around for a while, hasn't it?

Mills: The time has come for standards-based integration. Enterprises have always wanted to accomplish integration with information technology, but the evolution of the technology was such that they couldn't always realize the vision. Functional automation has always been possible but the lack of standards made it too difficult to integrate the functional pieces together. Now, we have enough standards to facilitate the interoperability of application function. This is where the big payoff is.

Tech Update: So, to summarize, you are saying that the bottom line should be the company's priority and that tight, front-to-back standards-based integration is a means to that end?

Mills: Look at Gartner's numbers. Forty percent of all IT dollars are spent on integration today. [Editor's note: Gartner cannot confirm this figure about IT spending on integration.] IBM looked at that in terms of both the money spent with vendors and the money spent internally on labor to make it work. We estimate that last year, the cumulative spend on integration alone was probably $1 trillion and too much of it was labor.

Tech Update: So, if integration standards can improve the bottom line by chipping away at the labor component, would you say that moving to standards-based technologies is a second priority for our audience?

Mills: The phenomenon of the Web should have been a crystallizing moment for everyone in terms of the value of integration standards. Suddenly, some very simple and readily accessible standards for the Internet like TCP/IP, HTTP, and HTML provided a core set of mechanisms that drove integration to new levels. That set off an enormous set of debates that raged in the early '90s and that had businesses very frustrated.

Think about it. On the heels of the '80s, when computing was all about decentralization, a client/server revolution came along that still left businesses with fragmented technology. It was fragmentation that nobody wanted. I mean, who would ever vote for fragmentation, right? You'd always vote for integration. But, at the same time you're going to want the sort of empowerment enabled by trends like decentralization and client/server, right? You're not going to want to run everything literally at headquarters. You need to empower your units, empower your people. You want to take advantage of what the technology as a tool can provide, but how do you get the balance of empowerment on one hand and integration of the business on the other? That was the dilemma that we were facing in the client/server era.

In the midst of all that, a social, technology, and standards phenomenon called the Web came along and proved that standards were the Holy Grail of integration. One of the greatest legacies of the dot-com era is the de facto agreement to a set of standards. It renewed enthusiasm on the part of both vendors and customers that standards can work. It came during a time when I was out there selling the stuff and I found a lot of cynicism about standards. Customers were telling me, "There's a lot of talk about standards, a lot of standards bodies, but I'm not feeling the effects of standards. I'm not getting the value that I thought I was going to get from all this standards work that has been going on." I don't hear that from customers anymore. I think the Web was an enormous catalyst.Tech Update: So the first priority is integration. The second is standards, but to the extent that it's a means to the integration end. You said there are a few other issues that might be nested underneath integration?

Mills: I truly believe that there is this one overarching priority and it's integration in the form of federation. It's not integration in the form of centralization. Letting the pieces be where they need to be, but tying them together and getting them to work together.

Tech Update: Federation?

Mills: You have federated data, federated transactions. You have federated human interaction. Collaboration becomes an integral part of every application scenario. I don't think of collaboration as a separate set of activities. Dating back to the client/server period, the things that you were doing from an individual perspective and within local groups often tended to be isolated within your group and your ability to create interactions that were broad-based across your institution and into other institutions were limited. The client/server phenomenon was in some respects a rehash of centralized approaches that were group-centralized but not institutionally-centralized. We went from massive centralization to small group-based centralization. In both instances the computing resources were isolated. Well, you now have the ability to create collaborations that extend across the entire institution that you're trying to automate and then between institutions as well. So you and I can collaborate.

Tech Update: When I interviewed Al Zollar, I remember him saying that collaboration can no longer be just an application. It has to be a feature of all applications. [Editor's note: Al Zollar is the general manager of IBM's Lotus Software Group and reports to Steve Mills].

Mills: It's a feature. It's everywhere. It's in the genes of every application. Except for a limited number of purely computationally-based activities, most commercial applications have a human interactive element associated with them. Think about designing things using computer-aided design. It requires profound amounts of collaboration. How do I collaborate with my suppliers and my sub-assembly providers? How do I get the best possible design, the best possible cost of manufacturing? And, I'm probably not vertically integrated any longer as a business; I use other suppliers or automotive companies or aerospace companies. I have suppliers by the thousands. How do we collaborate on getting the right design that gives us the best possible product at a maximum efficiency? Companies can no longer afford to have a serialized set of processes where I complete my design, hand it to you, you comment on it, tell me you don't like it, and hand it back to me, and so on. It's going to cost too much money to make it, change it, go back and redo it; all these time delays associated with serialization of interaction can now be compressed.

We can turn it into much more of a federated set of activities around human collaboration. We can collaborate on the design, we can get it all done right up front and we can speed up the process. There are things that we can do today with the Lotus portfolio and even more that we'll deliver a year from now for to better embed collaboration into any application or process.

Tech Update: So, in addition to integration, you see collaboration as a major contributor to the bottom line.

Mills: Acceleration. Velocity. Better return on assets. I improve my turnover rates. I get more efficiency out of my people. I get the job done faster. I can do more in a given period time. All those things equate to money in a business.

Tech Update: No mention of security?

Mills: There are a whole set of things then that get brought into the equation as characteristics of this environment. This sort of integration breaks down classic corporate boundaries and, as a result, you have security issues that have to be addressed. I want secure relationships with my partners, my suppliers. I need to ensure that I have those features built into my environment. When you're connecting all the pieces in these straight-through connected environments, you have to appropriately secure things that you don't want made visible to people on the outside. If you want your customers to self-serve, you want them to self-serve against the information that they're allowed to see. I need to secure the information about them--they don't want somebody else accessing their information. So, like collaboration, security is a feature. It's a part of the enivronment's DNA.Tech Update: Suppose I buy into your vision with respect to the bottom line and how standards, integration, and collaboration are a means to that end. Why should I turn to IBM to help me achieve that vision?

Mills: Solving these problems involves infrastructure. It involves the core building blocks of a system and so it involves databases, transaction processing, messaging technologies, systems management, security, collaboration, etc. These are the core infrastructure enablers. Applications take advantage of those things and need an infrastructure that's rich and robust and capable of providing cross-platform interoperability and connection--the base of this vision of integrated enterprises, if you will. That's what my team works on. That infrastructure. We don't build applications. We build the infrastructure.

Tech Update: So after you're done solving a company's integration problem, what products have you left behind? What's usually the prescription?

Mills: We show them how our technologies can tie their various applications together. Say you're running CRM, ERP, HRMS, design systems, procurement systems, and inventory systems. We use our technology to do a number of things for you. First, we can carry any one of those applications out to the Web in a secure, reliable, and scalable fashion.

Tech Update: Using?

Mills: WebSphere. Second, we can tie those applications to each other and get those applications talking to each other to actually pass the data and the actual context of the transaction itself across that entire value chain of your business. We do that with WebSphere, too. We can tie you to your supplier environments. We can allow you to participate in external markets and B2B exchanges regardless of their underlying technologies. We do that with WebSphere.

Tech Update: So, IBM recommends using WebSphere to bridge the old world to the new world, taking islands of technology and ...

Mills: Tying them together. WebSphere is the brand that we tie them together with.Tech Update: That's the middleware. Don't you need some sort of way to wrap those islands in something that WebSphere understands? You can't just plug them into WebSphere, can you?

Mills: We do that with a variety of things that are part of the WebSphere family. One of those is the WebSphere Studio Application Developer (WSAD). With WSAD, I can use Java, C, or C++ to wrap old things. I can use Java as a means to create connections between applications using XML. I can also wrap things and turn them into a Web service. I can do all of these things with the WebSphere toolset.

Tech Update: Does that include VisualAge?

Mills: WSAD includes the VisualAge programming environment. Through our acquisition of CrossWorlds,,we've recently added adapters to make it easy for developers to get prepackaged applications talking to each other very rapidly. So we have dozens of adapters for applications like SAP, Siebel, PeopleSoft, and J.D. Edwards. The CrossWorlds technology also handles industry-specific choreography of information flow. It helps to manage an order and carry that order through from one end of the business to another, all the while maintaining its integrity, ability to see where it currently resides in the process, and monitor its movements through my systems.

Tech Update: If standards-based integration is the Holy Grail, do the CrossWorld's adapters support the prevailing integration standards like XML?

Mills: Some of the CrossWorlds technology is XML-based and some of it is C-based. The world is not going to make everything XML anymore than the world is going to make everything Java. Traditional programming languages do certain things well. C is a traditional programming language. Java is a C-like programming environment that offers other advantages over native C programming. XML is yet another animal. It's a uniquely flexible tag-based language mechanism that is a very high-level structure that you can essentially use as a definition tool to define anything. Once something is defined in XML you can easily relate that definition to another XML definition. It's a mapping technique that often requires agreement to a standard. For example, a standard that defines all of the information associated with a financial transaction. Let's create an XML common normal form for all transactional information that's processed within financial institutions worldwide. If we could do that, then all financial institutions can surface those interfaces, all the application providers that sell into financial institutions can surface those interfaces, and therefore all applications can talk to each other within financial institutions reducing that massive amount of integration labor that we talked about earlier.

Tech Update: Well, if those standard definitions proliferate, why would I need the CrossWorld's special adapters? Theoretically, the need to wrap goes away, doesn't it?

Mills: The definitions definitely moves us from the hard-coded and more complex adapter mechanisms of the past into Web services-based interfaces of the future. But, the hard-wired adapters are a necessity to bridge the gap between legacy systems and provide a much needed jump-start for integration. So, where you lack those XML standard interfaces today, you can start to integrate applications with the CrossWorlds adapters. The goal is to eventually eliminate the hard-coded adapters over time and move to XML-based approaches. Once we get there, the choreography of the flows moves to the forefront, and that's where the CrossWorld acquisition takes on even greater value.