IBM's Linux Strategist Explains Gluecode Deal

IBM set tongues wagging in the application server community when it acquired Gluecode, an application server vendor that builds products atop the open-source Geronimo app server project managed by the Apache Group. Adam Jollans, chief Linux strategist at the IBM Software Group, sheds light on IBM's app server strategy, Gluecode's role alongside WebSphere and the SMB opportunities that Gluecode creates for IBM partners in an interview with Editor in Chief Michael Vizard.

CRN: IBM surprised a lot of people when it moved to buy Gluecode, an open-source application server provider, since IBM already has WebSphere. What's the thinking behind that move?

JOLLANS: The easiest way to look at this is in terms of our customer opportunities. We're seeing an increasing customer interest in open-source middleware. This doesn't mean that they're not interested in commercial middleware, but there is this emerging market for open-source middleware, especially in the application server space.

Given that we wanted to participate in the market, we then looked at a number of options. Should we, for example, open-source WebSphere, which is an option we looked at. And then we saw that there were actually quite a lot of open-source application servers. So we probably don't need another. Then the other factor was, where is the open collaborative community around these products, and where is it going to be? When we look at the really successful open-source products, it's Linux, Apache, Eclipse. And there's always an open collaborative community that everybody can take part in on a flat playing field of equal partners.

CRN: The other part of this deal that went largely unnoticed is that IBM will be contributing to the Apache Geronimo project from which Gluecode is derived. Is that right?

JOLLANS: That's broadly correct. We're supporting Apache Geronimo as an open-source application server, and then we're doing activities around packaging that up.

CRN: What's the core difference between Gluecode and WebSphere?

JOLLANS: We see Apache Geronimo as an entry-level J2EE application server. Some customers are going to be quite happy with that. Others are going to, over time, want to move up into the higher-transaction-rate application servers. We'll provide the ability for them to move up to WebSphere.

But there is no fixed dividing line between those, and actually over time it's going to move up. But the higher end is going to move up as well. Gluecode is going to be for customers who in the past haven't been using WebSphere, such as medium-size customers, departments, entry-level application and this whole concept of do-it-yourself application development.

CRN: IBM has tried to position WebSphere Express in many of those same markets. Does Gluecode overlap with WebSphere Express?

JOLLANS: Yes. But SMB, as you know, is a very wide market. Going forward, we'll see how this plays out.

CRN: When do you think Linux will go mainstream in the SMB space?

JOLLANS: I think it's application-dependent. With one of our programs, ISV Advantage For SMB, there are 200 hand-picked ISVs. We've now got about 17 of them with applications available on Linux. So now it's about how we help people adopt those and get those rolled out.

CRN: Is Gluecode going to get real support inside IBM, or will it simply be overwhelmed by all the resources being devoted to WebSphere?

JOLLANS: I think it will, because this is a new opportunity. This isn't about something instead of WebSphere. It's an incremental opportunity.

CRN: JBoss is widely recognized as a force in the open-source application server market. How does acquiring Gluecode change IBM's competitive stance toward JBoss?

JOLLANS: I think the difference we see is that Geronimo has got this open collaborative community around it. That means if you contribute stuff to Geronimo, then everybody can share. It's a level playing field in terms of all the players.

CRN: As you go forward, will IBM be moving features from WebSphere to Gluecode?

JOLLANS: We're looking at a number of possibilities. We want to see how the entry-level J2EE application server market plays out. CRN: Does this also give you another way to battle Microsoft's .Net at the lower end of the market?

JOLLANS: I think the key decision is down to, are you going to go for Java, straight J2EE or .Net? Now if you're going to make a decision for .Net, then you're going to go along the Microsoft line, using the application server they have as part of their products. If you decide you're going to go for the open, portable heterogeneous approach, then that's going to be the space of Java and J2EE. And there's a sort of an either/or decision. Geronimo runs on top of both Windows and Linux today. So this is an open-source play. It's not just a Linux play.

CRN: Microsoft has been talking about creating a unified object model by combining its file system with its relational database. You have donated your Cloudscape database to the Linux community, which is now called Project Derby. Will the Linux community try to bring Apache, Derby and Geronimo together to counter Microsoft?

JOLLANS: That's an option. I think it's too early to say that in terms of an incubator project at the moment. But we're using Cloudscape as a local data store in a lot of our products.

CRN: Speaking of Linux, it seems that IBM tends to favor Red Hat in the United States over other Linux distributions. Is there a reason for that, or is it just a general perception?

JOLLANS: We can always do more in terms of partners. But our basic strategy is that we feel it's important that there's choice in terms of Linux. From that point of view, it's good that there's more than one strong Linux distribution. Now, as an ISV, which is what we are as well, we look at the sort of cost of testing and supporting different distributions. We want to minimize the number of distributions that we're supporting for our products because this costs a lot and takes a lot of resources.

Remember, we're 350 products across three different chip architectures. You've got to minimize, which is why you come down to two as a good number in this space. It's not a perfect solution. But two is a good number to go with. We treat both Red hat and Novell equally in terms of partners we're working with.

CRN: That sounds like an argument to reduce the number of products and SKUs you need to support. Is that happening?

JOLLANS: We can try, yes. It's better than it was.

CRN: Given all the acquisitions IBM has made over the years in software--including Tivoli, Lotus and Rational--a lot of competitors would say that IBM's software group is akin to a Frankenstein project, where IBM is trying to tie a bunch a parts together to make something unnatural come alive. What's your response?

JOLLANS: There's a software group strategy organization, which is looking at those issues. And then there's a very big project where we're componentizing the individual brands. Instead of it being a WebSphere stack that includes Tivoli, security and a data store, what we're doing is componentizing the software into the individual bits like security or data store and then recombining those into products.

So you write the security once, you write the data store once and then you're able to use it in a whole variety of products, which is actually bringing the five brands much closer together. So you'll see a WebSphere portal combined with DB2, for example. We've done some revamping. For example, DB2 Information Integrator has become WebSphere Information Integrator. We'll also transfer ownership into different brands, so WebSphere portal is now being developed by the same team that is developing Lotus Workplace in Lotus. We believe componentization is a better way to build complex software.