John Swainson is general manager of IBMs Application Integration Middleware division, and is also known as a pioneer of the WebSphere application server. Version 5.0 of the server was announced last month, when Senior Editor Darryl Taft caught up with Swainson in Boston for an exclusive interview to talk about the new product, where it puts IBM competitively and whats to come in the application server software space.
EWEEK: Whats your role at IBM?
SWAINSON: Im the general manager of a unit called Application Integration Middleware. Its one of four major software product divisions at IBM. And I have responsibility for the development, marketing and sales of our middleware products, including the WebSphere family of products, application development tools, MQSeries, communications products, transaction processing products like CICS and other things. Basically, all the tools that we build for developers to use and a lot of the things that sit above the operating system and below the applications.
EWEEK: Was it you who built the WebSphere business into what it is today?
SWAINSON: Ive been in this role since it was created in 1997. In the last 10 years Ive principally been responsible for the development of our application development tool products. I was the lab director in Toronto. And I had responsibility for the business unit that included application development tools. Then we merged that with our business unit that did transaction processing and we announced WebSphere out of that in early 1998. And Ive been the general manager of that group since that period of time.
EWEEK: How much does WebSphere contribute to the bottom line?
SWAINSON: Well, we dont break it out exactly. IDC and others we think have a pretty close estimate of the total amount of revenue that WebSphere app server and the total WebSphere family contributes. In round numbers the app server is a little less than half a billion dollars. And the total family is around a billion. The middleware software business in total this year is about 11 and a half billion dollars. So the WebSphere brand amounts to about 10 percent of that—rounding up a couple of digits.
Page Two
: IBMs Swainson: WebSphere and Beyond”>
EWEEK: What are the highlights of WebSphere 5?
SWAINSON: The first priority on WebSphere version 5 has been performance and quality. There are lots of features in WebSphere version 5 including J2EE [Java 2 Enterprise Edition] 1.3 support, a lot of the Web services standards have been rolled into this release. We have support for some additional platforms–some of the Linuxes and other things. There are quite a few significant technologies in this, but the big picture view of this is focused on reliability, stability, performance, quality and tighter integration between the application development tools and the runtime environments. What weve been doing with Eclipse shows where the WebSphere application development tools have superseded the previous generation of Visual Age tools and have been tightly integrated with the various runtime environments they support.
EWEEK: How does WebSphere 5 stand up against the competition?
SWAINSON: Its the building block for all of our WebSphere family offerings. So thats obviously incredibly significant. It takes things like performance and scalability and reliability and ease of maintenance and ease of operations to a new level. So customers who deploy this in large-scale environments like eBay are going to find it easier to maintain, easier to operate and easier to use. That I think will provide us with a point of competitive differentiation. We still have the fullest range of offerings, all the way from Intel with Windows and Linux, all the way up to the mainframe. It supports very high performance, very high scalability applications that run thousands of transactions a second in a full transaction processing kind of environment. These are things that set us up against the competition.
EWEEK: Can you talk a little more about you small to mid-market strategy?
SWAINSON: The small and medium business market for us, particularly the medium business market, is very important. Its about half of the total middleware market, we think, with literally tens of thousands of customers who buy infrastructure technology. There are a couple of characteristics to this. One is that it tends to be a market thats driven by channels. Often ISVs and integrators are the principal method of going for this market. So your products for this have to be very channel friendly and partner friendly. In general the demands of the SMB marketplace are simpler in the sense that theyre not going to be driving thousands of transactions a second. On the other hand theyre going to value ease of installation, ease of operation and what we call near zero administration—really being flexible and easy for people who dont have large-scale IT departments to really deal with.
EWEEK: It seems that not only in the small and medium business markets but overall that people are expecting to pay less for the core features.
SWAINSON: Theres no question about the fact that there is a base level of functionality that people are coming to expect from their products. And yet customers demands are at the same time pushing us into higher levels of functionality. So for people who want to do simple things you have to deliver simple products that are appropriately priced. I will tell you that there are large numbers of customers who are pushing the envelope in terms of scalability, performance, and reliability. Lots and lots of people who are using these things in 7 by 24 by 365 kinds of contexts for whom the issue is not necessarily low purchase price, its low operations price and its a high degree of integrity and reliability over the long term.
Page Three
: IBMs Swainson: WebSphere and Beyond”>
EWEEK: BEA Systems has knocked IBM on not having J2EE 1.3 support, they claim your development tools are complex and their claim is that even with WebSphere 5 youre going to be six months behind them.
SWAINSON: I would remind you that we GAd [made generally available] our first J2EE 1.3 release in December of last year. We have in fact had a J2EE product in the market for over a year. And unlike them we actually tell people when we ship things that they can use them. BEA will say its available but we dont recommend using it because we havent tested it. So our philosophy here is very different and it really comes from our heritage of building large scale, high-volume transaction processing systems like IMS and CICS—which are still the things that run almost all the large banks and insurance companies and airlines in the world.
I frankly dont like to talk much about BEA. I think BEA spends all their time talking about IBM and none of their time talking about how their products help customers and I would much rather spend my time talking about how Im working with my customers to allow them to get business benefit rather than how much better I am than BEA or how much worse BEA is than somebody else. I just dont subscribe to that business philosophy. It seems to me counter-productive.
I think BEA has a credible product and I dont know why they dont focus on their products. It amazes me that they want to spend all of their time talking about IBM.
EWEEK: But philosophy aside, how do you compete with the various vendors in your space, including BEA, Microsoft, Oracle and Sun?
SWAINSON: Let me start at the macro level. There are two schools of thought here in the marketplace. One is based on open standards. And a number of us, including BEA, fall into that category. And then Microsoft sort of lives in the world of proprietary standards. They argue they are open standards, but the reality is the implementations are mostly proprietary. So we compete largely on a standards basis with Microsoft. So if a customer decides they want to go down an open standards path, then you segment the competition into ourselves and BEA and Oracle and sometimes iPlanet [Sun ONE]. If they dont, then you have a fight with Microsoft around .Net implementations.
We compete with BEA typically around things like scalability and performance, and application development tools. We see more of BEA than anyone else. In most of the enterprises that we compete, BEA is at least a factor—whether its an installed WebSphere account or an installed WebLogic account, most of these customers in fact have both. Most have chosen to have both.
We see Oracle typically in situations where Oracle is the incumbent from a database perspective. Oracles sales pitch appears to be were going to bundle this with our database so you might as well use it. And we almost never see iPlanet (Sun ONE) even on the Sun platform ironically. Most of the time the Sun sales force continues to sell with BEA where the customer has chosen WebSphere they sell with WebSphere. So we see less and less iPlanet in the marketplace overall.
Page Four
: IBMs Swainson: WebSphere and Beyond”>
EWEEK: How have IBMs acquisitions helped you in your role?
SWAINSON: Our strategy is to be opportunistically looking for high-quality companies with high-quality technologies that allow us to augment our product portfolio. The Crosswolds acquisition got us really firmly into the business process integration space, which we hadnt had an offering in before. We augmented that with Holosofx because we needed a tool for business process integration and Holosofx had a very good one. Wed had a long business relationship with them. Were pretty focused on filling out our portfolio.
EWEEK: Why does IBM take the approach that the application server is the right architecture for integration?
SWAINSON: Well, the app server is a foundational architecture for a lot of different things. We have a different message around integration than a BEA who talks about their build-to-integrate strategy. We fully agree that you can build applications to be integrated. But in many cases there is just as valid a case to be made for integration through a database. And a lot of applications get integrated through some sort of federated data scheme. Obviously data warehouses are in fact just a way of doing that. Another very valid scheme for integration is to use a message-driven middleware like MQSeries to tie applications together or use a message broker like the WebSphere MQ Broker. So we have a rich portfolio of integration technologies that include the app server, but we dont limit our message to being the app server is the only way to integrate.
EWEEK: Does IBM see Web services as synonymous with app servers?
SWAINSON: No. Theres a high degree of correlation in the sense that many Web services will be delivered by the application server but not all. So, for example, the Web Services Security architecture that were working on with Microsoft, some of that will be delivered by the directory. Some of it will be delivered by an authentication service that can be independent of an application server. Things that are directly related to applications, like the Web services transaction protocols and that kind of stuff, thats obviously going to be part of an app server. And up to this point most of the Web services standards have been around applications and therefore weve delivered them through the WebSphere app server. But thats not always going to be the case. Theres a whole set of systems management standards that are evolving here. And systems management stuff wont all be delivered through the application server.
EWEEK: Do you think J2EE should have to create new APIs for every new Web services spec like theyve been doing? Or should there be another way they can go about it?
SWAINSON: This is a religious discussion too. The answer is not necessarily and probably not exclusively. We believe that Web services need to be independent of the implementation. So you need to be able to define a Web service whether something is written in J2EE or COBOL or even .Net. And therefore to try to turn this into something that is exclusively the province of Java or J2EE we think is a mistake. We are a big supporter of J2EE and continue to be a big supporter of J2EE. But the world is not just J2EE any more than the world is just .Net. Theres lots of COBOL out there, theres lots of C and C++.
EWEEK: Another argument against IBM is youre the complexity company and primarily you want to sell services. How do you respond to that?
SWAINSON: My business has nothing to do with selling services. My business is selling software and selling software to customers whether its delivered with services or not. Having said that, of course we are interested in solving the complex problems that tend to go along with the worlds largest enterprises. So if the accusation is that IBM likes solving complex problems then we stand guilty as accused. However, we also solve simple problems, but our core customer set is the enterprise market. And the enterprise market tends to have more complex problems than anyone else. There is no tie at all between IGS [IBM Global Services] and our software business.
EWEEK: Also, competitors talk about the size of WebSphere—that its a hodgepodge of products.
SWAINSON: This is another one of these broad overarching statements that is not true. The WebSphere family is not one product; its a family of products. It consists of three major elements: a set of products to build the applications—the WebSphere application server and associated tools; a set of products to deliver those applications—things like our portal and our commerce product—and a set of things to integrate those applications into a business. That seems like a pretty simple way to view a family of middleware products to me.
EWEEK: What can you say about the future of the business?
SWAINSON: More customers today are trying to worry about how they integrate the last 30 or 40 years of building applications than they are necessarily worrying about how to build the next generation of applications. And so this whole area of business integration and business process integration has become much more top of mind for customers and much more top of mind for CIOs as they try to make their businesses more efficient and more effective. So I think you will see from us a great deal of emphasis on business integration, business process integration, tools that make it easier to do that and faster to do that. By some measurements customers spend as much as 40 percent of their total budget on integrating their applications and we think we can provide some significant benefits through automation to that process. So I think youll see a lot of emphasis on business integration and business process automation in the medium term.
Long term I think youll see a move to service-oriented architectures and Web services with everything being thought of in the context of a service and applications being composed out of services. So very long-term I think youre going to see application development being done differently than it is today. I think youll see most application vendors shift to a more standards-based way of delivering applications. Their applications will have standards-based interfaces on top of those such that they can be composed into new applications by the customer or by systems integrators. Thats going to have some interesting effects on software pricing models that I dont think we fully understand yet.