Think Tank: Enterprises Open Doors For Open-Source Software

Optimize and InformationWeek editors jointly held a roundtable discussion on the topic of open-source software at the recent InformationWeek Spring conference in Hollywood, Fla., to get a sense of how viable a solution it is for enterprise users. Participating were: Raven Zachary, director, Internet technology, La Quinta Corp.; Pedro Villalba, senior VP of information technology, HIP Health Plans; Richard Plane, chief technologist for information services, Harris Corp.; James Hatch, principal, The Hatch Group; and Ben Sabrin, director of North American sales, JBoss. The discussion is presented here in an edited form.

Optimize: What's your interest in or experience with open-source software?

Raven Zachary, director, Internet technology, La Quinta Corp.: We've been using open source at La Quinta for two to three years now. I also used open source prior to coming to La Quinta, at Excite—before its bankruptcy, but not the reason for its bankruptcy; let me make that very clear. La Quinta.com is powered by open source: Apache Web server, Tomcat, and JBoss, and we rely on an Informix legacy database at the back end.

Pedro Villalba, senior VP of information technology, HIP Health Plans: At HIP Health Plans, I'm looking at open source from a different view, more like two or three years down the road. As I see the costs of licenses and operating systems escalating, I look for alternatives to deliver quicker solutions for the business functions—which still seems to be the biggest cog in the wheel. In our environment, the speed of the infrastructure is almost doubling every two years, and business executives are becoming well-educated about what's available in terms of different devices and the speeds at which they can get information. But the application-development cycle is still long, and there's a lot of pent-up demand to speed the delivery of functionality to keep pace with the infrastructure. However, application-development costs using proprietary software aren't especially low, so I'm looking at various options. The biggest concern we have about open source is its level of accountability.

Richard Plane, CTO, Harris Corp.: Harris is a $2.5 billion company, and we're somewhat of a pioneer in utility computing. We run a for-fee J2EE platform and I'm wrestling with open source on a couple of fronts. One is an ERP system that is moving off of proprietary software onto JBoss. We had to figure out a way to support that as well as putting in redundant arrays of nodes in the development process. Like Pedro, application development is where a lot of my cost comes from because I need copies of databases for the developers. By comparison, production is pretty much under control. So, if I can move my development onto this new platform and lower my costs, I think I can be very competitive with India and Chinese development costs.

James Hatch, principal, The Hatch Group: I've been a CIO in many different companies, and a couple of years ago I started a small consulting firm. I coach CIOs and CXOs in better managing the whole issue of information processing. I attribute my interest in open source to my "James Dean school of management" attitude where I rebel against the existing establishment. But there's a more pragmatic reason, too: Three or four years ago, I came to the realization that the IT world needs a far more cost-effective mechanism for the desktop than Microsoft. Open source is a great improvement, not only on the cost side, but also in practical functionality.

Ben Sabrin, director of sales, JBoss: I've been with JBoss for two years now. I started as the first nontechnical employee and defined our business model and our growth strategies, and educated people on open-source licenses and strategies. JBoss is an open-source player for J2EE application services.

Optimize: Is open source now mainstream? Is Linux? Are we at the second phase of open-source initiatives? See sidebar: Survey: Open Source Finds Opening.

Plane: I think we are at the next phase. I've been in IT for almost 25 years, and I can think back to when Sun came on the scene. It was the open-source leader. The network was the computer; everything was open source. I think Sun has suffered the first blow.

All of a sudden, you can take Linux and lay it on this Intel platform and completely take any of Sun's value out of the game. Phase two starts with Microsoft's presence on the desktop, scrambling with its 2003 server office system, trying to identify the value of having this fully integrated suite. As James said, you start to wonder what their value proposition is.

Hatch: Yes, open source is an established fact, and we are entering a second phase. It's an established technology in the server world. And it will go to the next level, but it's not there on the desktop yet.

Villalba: I think we're in the second phase because inherent in the products you buy is open-source code. We bought some Avaya products that run our call centers; Cisco now has some products with Linux. So you're buying these products indirectly. You know that's what's making that equipment run, providing that service. I've made a big point of making my executive team aware that it's coming.

Optimize: Would you say Linux is mainstream, then?

Plane: When vendors like PeopleSoft start to support it, it's mainstream in the server infrastructures.

Zachary: I'm fascinated that when you ask the open-source question, people immediately jump to Linux. In my world, I don't even have Linux. I consider La Quinta open source in terms of the online initiative. We're running Solaris as our OS primarily because we think it's a better OS for Java. There's a practical reason why we don't have Linux in play. But I think of open source in terms of things like Apache Web server and a lot of the Web services and SOAP [Simple Object Access Protocol] initiatives that IBM and Apache report. And open source is fully mainstream in that area. Apache has been the No. 1 Web server since the inception of the Web.

Sabrin: Linux has gone mainstream through IBM, Hewlett-Packard, and every other very large vendor behind it. IBM had a Superbowl commercial with Mohammed Ali talking about Linux; it doesn't get more mainstream than that.

Optimize: Pedro might not be quite convinced, yet.

Villalba: No, I'm not. Coming from a $3 billion company processing health-care transactions and laboratory-test results that a physician is waiting for while he has somebody open on the table, I have different concerns.

We don't have our heads in the sand, hoping that a Microsoft product doesn't go down; we're starting to play with open source in the development area. We've set up laboratories for our development staff because the number of server boxes they need to develop future applications is overwhelming my budget.

So I've allowed for some open-source testing to take a certain infrastructure and slice it up so multiple people can work on a development piece. I'm just playing today, but no doubt in two to three years you'll see [open source] taking shape, especially as we get more health-care and insurance companies involved in code developed around specific applications. I'm investing R&D dollars now because I want to be ready in three or four years to have enough trained resources, so that when it hits the floor, I can start moving some mission-critical application to these products.

Optimize: Is that R&D development that never sees tangible product use, or is it actually being used in test-pilot situations?

Villalba: We're actually using it to collect information just as we used wireless-PC tablets about a year ago, but not for decision-support applications.

Optimize: Do you see some negativity in the culture of open source zealots, perhaps? Do you think it's muddying the perception for everybody else?

Hatch: There's emerging what I call an open-source culture that suggests that all the really neat stuff coming down the road is going to be open source. I guess that's more a wish than reality, but you can see that creeping into the conversation that the best products quite likely will be open source.

I suspect it's partly the developers; they really have the mentality that "I get my jollies out of sharing my creative work with others who can then take it to the next level, and collectively, we're going to get somewhere faster because I'm not constrained by all that bureaucracy."

Sabrin: In security, you have the hobbyist, the hackers, versus the people who do it for pay, and the hobbyists always win. So a lot of what's going on in the open-source community—at least what we're doing—is reaching out to those hobbyists and giving them a means to make a full-time living. But you still have the huge peer review, so there's this insanely intense review and the ability to strive to be the best.

Hatch: When the software world started, the first real open source was IBM's mainframe operating system. IBM made the machine and said, "The software is free; we give it to you, and you guys all do your improvements on it in your shop." And there was a collective spirit in the IT community in the '60s where you wanted to be known for the improvement you contributed to the operating system.

Plane: It was the same for the Web. It drove us from the way we were getting information through FTP and wide-area search engines—all the things we had in our tool kit as researchers we thought were cool—where these guys come up with this whole notion of graphical presentation. It's the same kind of thing. The collective minds of the world, focused properly, are a wonderful thing, and it's the few bad apples that can ruin it.

Optimize: Do you agree that open source's key benefits are lower costs, openness, and accessibility.

Plane: Well, the real value is the pace of innovation, not the cost. I think we'll see far more innovation without the shackles of quarterly stock valuation from the publicly traded companies like Microsoft. They do things for profit. Open source doesn't necessarily live by those rules—though that could change when you start to commercialize it. They have this worldwide consortium of people putting modifications in; they're just dealing with release control at this point. So the pace of innovation is key.

Optimize: How do you sell this concept to your top management. They're concerned about legal issues, about privacy. What do you emphasize?

Villalba: We are primarily a Microsoft shop. Six years ago we didn't even have any Microsoft products or any desktop, so we had to introduce it and millions of dollars were invested. We did a good job convincing the management team then that it was the right investment. We're doing a pretty good job convincing them now that we have to look at open source, too, that this is the future. We use those terms to make them feel comfortable so they can give us the millions of dollars we need to build that infrastructure and so our client/server application developers can play in that sandbox.

Their only concern is the legal issue. They know that we're running a well-managed IT shop at the right price point in terms of revenue that we collect; we do a lot of best-practice market analysis. So they're not looking for me to lower my IT cost. But what I sell them on is this: "This is the future. This is the way we're going to keep up as you double the amount of insurance products you want to sell." I'm making the point that I need to explore this; it's the only way I'm going to be able to keep up. Then the other question they ask is reliability—is it industrial-strength?

Hatch: How could you put it on servers if it weren't reliable?

Villalba: Well, this is the executive level. This is a president, a CEO, saying, "We trust you. We've made an $80 million investment just this year in IT. But is it going to be industrial-strength? Is it going to take us to the next level?" That's what they want to know.

Optimize: How does La Quinta deal with the legal issues?

Zachary: Well, most of the legal questions have related to SCO and Linux. It's a nonissue for us; we're running Solaris. But that's not the extent of the legal problem. In my opinion, the legal issue in open source is no different than two closed-source companies suing each other over intellectual-property (IP) issues. We'd have the same problem if PeopleSoft sued BEA.

Sabrin: If there's legitimacy to the SCO lawsuit, it's from a purely IP standpoint. Open source doesn't equal lack of IP in our mind. We don't think IP disappears. We just think the license that you grant to use that IP changes.

Optimize: Since many off-the-shelf enterprise products already have open-source code, you're inheriting it indirectly in some cases. Any thoughts about that?

Villalba: A lot of infrastructure technology, call-center technology, etc., has open-source code. So we've educated executives in the benefits of purchasing and acquiring products that have some of these operating systems because that allows us to do hybrid implementations.

There, too, it comes back to the legal question. I had to respond to an E-mail from our president because he read an article and he asked: "What does this mean to our Avaya product that we just bought?" because we know it was Linux-based. I said I think we're OK, because if they're going to go after someone, it's Avaya.

Plane: I've had a similar experience with Google and Avaya. These things just don't go down. They've got embedded Linux; they run 24 hours a day. That's going to help you educate the management team. You can say, "This is rock-solid; we've been running the voice circuits over proprietary suites onto this open-source platform and it hasn't gone down for 15 years."

Villalba: Exactly. That was the strategy. They trust the name Avaya now. It's very reliable, and open source isn't an issue.

Optimize: How about total cost of ownership? Once you build the in-house expertise and manage these things, and you pay for the hardware, are you really saving money? Plenty of reports show that Windows' TCO is lower.

Sabrin: I'm not going to disagree that there are multiple points you need to address from an open-source stack perspective. And at least with regard to Linux—not necessarily Red Hat, but enterprise Linux in general—the server is starting to blur some of the TCO advantages. At $3,000 or $2,400 per server, that's not as attractive as it once was. Now, granted, you're still free to use that source or use it in a package, and you could still get the cost advantage there. The problem is that commercial-software companies are certified in their product only for the enterprise lines, and customers can't deploy free Linux anymore. They have to deploy commercial Linux, which is another turn-off for me.

Plane: I have a lot of transactions, so I have very big Sun iron back in the data tier. Once you've gotten to the Java platform, you can start to consider redundant arrays or inexpensive servers, and you can stretch this out on quad NT boxes and get comparable performance at 10 cents or 20 cents on the dollar.

Optimize: Another criticism of open source is ease of use. Currently, if you want to use an open-source stack, you go to MySQL for the database, JBoss for the app server, Red Hat for the operating system. It seems like a lot of work.

Plane: You still have to distribute the software and do a lot of things that Microsoft, by the way, has come a long way with. If you're running IT like a business, as I am, and you're driving at 1% of revenue, when you're at that level of efficiency, it's hard to start seeing the dollar savings.

Optimize: Do people look at their staffs and say, I can't do open source because I don't have the people?

Plane: It depends on how you categorize open source. If you look at it from Linux to Unix, it's the same. Guys who do Unix can do this stuff with their eyes closed. Java is a beast. You've got to work on Java to get it to be stable and running enterprise applications, but once you get past that, the barrier to buy BEA versus JBoss is low.

Optimize: Do you do a lot of training or retraining?

Zachary: No. I had the luxury of being the first team member in, and I hired a certain skill set and level of competence in open source. This is a great question, though. How do you take a C++, .Net, C# group like you have at HIP, and say, "Let's go Java." That's a big issue—converting a nonopen-source development team to open source.

Plane: That's the struggle I have. All my Web presentations are Microsoft. I have to retool everybody, and they're schooled in Visual Basic, and they're skilled in the .Net.

Hatch: Moving those C++ people to an open environment is a killer.

Villalba: It's a challenge if you work with client/server people, which I do. We put in $2 million for these young guys—they're 30 to 40. You're not going to keep these good employees unless you let them play with some of this stuff—and if you think they're not playing with it, they are. You're not going to keep it out of the shop, so the key is to channel that energy, provide them a park where they can play and I can watch them.

Sabrin: It's a different culture mostly because the approval process associated with the acquisition of new tools is completely taken out of the equation. In the open-source world, if somebody wants to go get JBoss, they download it. There's no procurement process—or costs—associated with that acquisition. That kind of touches on the whole TCO argument, and I believe has a favorable impact on TCO: You eliminate the procurement costs associated with technology in your organization.

Optimize: Another key issue is the idea of a heterogeneous environment. Do you see commercial software and open-source software working together

Plane: Yes. I run a PeopleSoft HR system, although it's fully Web-enabled and it's running BEA version 5.1. I want all of the functionality that's been released by the Java open-source group that BEA slowly funnels in, and that may be where I make the choice. I may have the for-fee version that's delivered to me from the software supplier. But the advanced presentation, advanced-query ability, and advanced business-modeling process are all available in the new versions of JBoss, so I may want to run that side by side and take advantage of that functionality.

Now the noodle I have to get around is taking my .Net developers into an Apache environment, and that's really a tough one to break down because I already have that embedded skill set.

Sabrin: I see heterogeneous environments every day, and maybe the thing that can solve the problem, and that will alleviate having to retrain those client-side Windows developers, is Web services. I have a Java run time that's proven to be more scalable and robust than what you have from .Net, but I don't have to change the way my client-side developers think, right?

Plane: You can use Web services as a bridge between applications and remove a barrier to entry for the Microsoft approach.

Optimize: Where do you expect open source to be in two years?

Sabrin: If you just look at the Internet, and you think about your operating system, database, application server, Web server, that stack of software is going to be open source. A lot of that will be driven by not only what's going on in the open-source software community, but by hardware vendors. How are HP and Dell going to get a dollar out of a commodity hardware box? If they're going to find margin, it's going to be through an open-source software stack that has some kind of managed service associated with it. Proprietary software is never going to die, provided there's significant value.

And open source will help push more innovation at proprietary-software companies. We're never going to write a business-intelligence or an ERP system. But we are going to continue to push the layer of innovation down to the system level instead of building abstract interfaces on top of this infrastructure.

Villalba: I'm hoping open source will allow me to get more business functionality out the door faster. That means application development can keep up with the infrastructure and the fact that every six months our company is putting out four to six new health-insurance products. Whether it's a combination of Web services or some other tool that will allow me to put those out there quickly, I'll ratchet up to make them industrial-strength.

Plane: I don't think the guys who have shareholder value are going to throw in the towel. We're already starting to see Microsoft targeting open source, and Sun targeting open source, instead of each other. I don't know whether we'll get to some kind of hybrid between what's in the open-source space and what these other guys provide, but I think we'll start seeing innovation.

We're going to see more compute cycles for service, and that will be hosted on open source over IP because of the cost structures. So in two years we're going to see more moves away from the notion of client/server, more into network computing. That's what I'm moving my business toward.

Hatch: I'd offer one caution: The transition from what I call the new legacy—which is all the Solaris-based apps, all the HPUX apps, all the old mainframe apps that are still out there running large corporations—will take a lot longer than people imagine. Half the apps out in the world are still Cobol. They don't go away very easily. You've got to recognize that. I like the enthusiasm that says, "Man, we're going to get there in two years, and this is going to be open source in two years," but it won't happen that quickly.

Zachary: Two years isn't long. In the OS area, I think you'll see more focus on horizontal scale. You're going to see really cheap, small Intel-based servers. That's what we're using. We love them because we can run Solaris today, and we can reformat them tomorrow, and they're cheap. They work great.

In the next two years, the most dramatic and most visible change will be people reformatting Windows servers in a data center with Linux because it's simple and it's cheap, and they can do it. On app servers, it's battleground time. It's BEA versus IBM versus Sun versus JBoss, and in the next two years that battle will be fairly noticeable. In the database area, [open source] isn't there yet. You've got players like MySQL, but they're not ready to take on open source for a long time.

Please send comments on this article to [email protected].