IBM Exec Defends Proprietary Middleware on Linux

One of the paradoxes of the industry is that while IBM is an ardent supporter of Linux as an open-source operating system, it has a more proprietary approach to the middleware given its history with Websphere and other products. But IBM argues that as open source continues to evolve, the company will be contributing more middleware to the open source movement. In an interview with CRN Editor In Chief Michael Vizard, Jeff Smith, IBM's vice president for Linux middleware, explains the company's evolutionary approach to open-source development.

CRN: How do you explain the apparent disconnect between IBM's support for Linux in contrast to its vast portfolio of proprietary middleware products? Some people would say that IBM is hypocritical, because while you endorse open source at an operating system level, the company continues to make billions of dollars of proprietary middleware?

SMITH: First of all, I think that statement in itself is hypocritical. Whether or not something is open source or not is really not even the question. We support an operating system that is based on open standards. It also happens to be open-source project. We need our stack of middleware to support our customers' business applications. There's nothing inconsistent or hypocritical about that combination because it's all focused on what value the customers need to solve their business problems. If you look at the IBM middleware portfolio, we believe that there's a tremendous opportunity for all of that stuff as the whole adoption of Linux for business workloads continues to expand.

CRN: What's your take on companies like JBoss and MySQL who are competing with IBM on Linux platforms?

SMITH: If you look at JBoss they are not open projects like Linux or Apache. They are commercial ventures that have allowed certain types of controlled access to their code, and they have tried out a different pricing model. Their approach is to make the acquisition costs of the license really cheap or free. And then they make all the money by selling the version that really will run your enterprise workload. It's not a terribly bad idea, and it might have an interesting opportunity for certain marketplaces. But at the end of the day you've got to ask what is its flexibility? What is its reliability? What is its security? What is its scalability? What is the actual ongoing total cost of ownership? And what is its appropriateness for what you're trying to get it to do? We think that for huge amounts of what people want to do with middleware, our stuff is what they need and that MySQL and JBoss will be insufficient, independent of any questions about whether something is open source or not.

CRN: What kind of workloads really favor IBM in that equation?

SMITH: If you're talking about somebody that wants to take a single, small server, fire up a database that an application can put some information into it occasionally store and retrieve stuff from it, it doesn't need to scale across lots of boxes, it doesn't need to scale thousands or millions of transactions in a short period of time. And it isn't underneath a mission critical workload that costs lots of dollars if they go down. If all that stuff's not so important to you, you just kind of need a regular old database or you just kind of need a regular old application server, and then there are lots of choices for that. In fact, if you really just want something that's truly open-source that does a basic level of Web serving, then what we would say is just go and get Apache and Tomcat and do it yourself. But the additional capability around stuff like MySQL and JBoss takes you a little bit further, but not nearly as far as you would get with the data management portfolio you get around DB2 and Websphere.

CRN: Is IBM actively contributing middleware to things like the Apache and Tomcat projects?

SMITH: Absolutely. We are major contributors to Apache and have a probably less direct involvement in Tomcat. We also recently put 500 patents out into the open-source community. And we had a technology that we called Cloudscape, which we put into Apache as part of their open-source project. That project is now called Derby. We'll continue to expand the value of their technology and the open-source stuff will continue to slowly move up the stack. And as that happens then the folks like IBM and the commercial middleware business will continue to move our value proposition up the stack. It's a sort of natural progression for technology that has been occurring since the beginning of technology.

CRN: Microsoft would argue that bundling a lot of the middleware functions into the operating system, they have a total lower cost of ownership. What's your take on that argument?

SMITH: At the end of the day, I think a lot of customers don't care so much about that. The bottom line is that they know that to solve a business problem they've got to have a piece of hardware, they've got to have an operating system, they've got to have some middleware and they've got to have an application and then they just want that stuff to sort of work together as well as possible and support their ultimate goal. Customers like to be able to think about all that stuff as a solution. They expect the total cost, both to acquire it and the ongoing cost of ownership for that stack, to be something impinging on their business model. But at the end of the day, you know which pieces are part of the OS and which pieces are middleware is not a debate that is particularly interesting.

CRN: What kind of traction do you think Linux will have in the SMB marketplace?

SMITH: Midsize businesses are an interesting place for Linux because most midsize businesses are served by some kind of a business partner. Huge portions of this midmarket are served by these guys. And these guys love Linux because there's a lot of flexibility in how you can acquire it and build it into solutions that you might put forward as a system integrator. It's easy to build services around, which is the way these guys like to make money. It's an easy, flexible way to provide a solution without some of the handcuffs that come with building your solution around some of the more proprietary solutions out there.

CRN: Other concerns that get tossed around is that Linux is supported by an army of nonprofessional hobbyists. What's the real support infrastructure around Linux?

SMITH: If you really kind of look at how the support structure for Linux gets executed, particularly if you do business with IBM, you're not going to have much of a clue that it's any different than what you were used to. You got a single point of contact to come into. We've got folks that do initial diagnosis, and triage all that kind of stuff. And in working together with the open-source community, they even have the capability to deliver you a fix to the Linux operating system that'll get you up and going and pass your problem. So from a customer perspective, it is every bit as rock solid as what they are used to on something like AIX. It really works quite well.

CRN: Beyond the fact that Linux is free, why do you think Linux is becoming popular?

SMITH: If you start listing those reasons you see that it's really flexible in that it runs across lots of different kinds of hardware. It's easy to put lots of different kinds of software on top of. It's highly secure and there are no secret hidden back doors or anything like that. It's got tremendous reliability characteristics, not only because of its Linux heritage, but frankly because it's been around a long time. And it's less expensive to own and operate. The fact that it's open-source is also interesting and valuable, but that's not really the main point.