Virtusa, an IT services company founded in 1996, was using proprietary version control and collaboration systems to develop software for its clients until Sri Lankan founder Kris Canekeratne decided that a custom solution built on open source components was a better fit for internal use. As a result, the company ended up saving millions of dollars on licensing fees and acquisition costs.

It was as a satisfied consumer of the Drupal content management system that Bachana first realized the business success potential of moving away from proprietary licensing structures. “When we made a decision to do more interactivity on our Web site, we determined that it made the most sense for us to implement a content management system on an open source platform.” Bachana began hiring what he calls “open source gurus,” and he found their enthusiasm for community-based development contagious. “They evangelized within the company about the merits of open source. Having some new people in this environment, which had always been traditionally Microsoft-based development, well, they were sort of getting people excited about what could happen.”

It didn’t take Bachana long to connect the benefits of open source content management, such as drastically reduced capital requirements and greater flexibility, to meeting the needs of his clients. “I get really excited about solving business challenges,” he says, calling consulting a “buffet lifestyle. You get to solve challenges in lots of different businesses. We found a number of our customers didn’t have the budget to purchase licensed products. In university settings, or museum associations, or even more recently media companies, there’s been some issues around decelerating of their revenues. They just couldn’t afford [proprietary]. For us it was a logical next step to offer those customers open source.”

Bachana says DPCI also began getting requests from clients specifically for open source solutions. “They asked us to go out and recommend platforms in the open source world that we could help them implement and customize. When we first started nine years ago, we were either building custom solutions from scratch, or we were implementing proprietary solutions from big-name companies.”

Moving to open source inside and outside the company has benefited DPCI in more ways than one. “Rapid deployment,” Bachana says, is the biggest benefit. “Not only within the framework of Drupal, but other open source modules and components that we can integrate. And we work in a LAMP environment, so it is quick for us to implement new functionality that we want internally.”

Using open source software also benefits DPCI from a business strategy perspective. “One of the drawbacks of consultancies is that there are typically a lot of solo practitioners that can’t do the bigger projects. We use a team approach, and by matching this with open source, we think it gives us a strategic advantage, because the team can implement changes very quickly [for] our customers. That’s been a terrific benefit for us.”

Bachana says the challenge in open source is putting all the pieces together. “The market is not fully mature,” he says. “There are pieces to the puzzle that we’d like to see, like customer relationship management, accounting — all the different pieces you’d see in managing a business. There’s still a lot of satellite initiatives that haven’t converged yet. The disconnect is that there’s no one entity or group or central place where people are thinking about how all the pieces snap together. I’d like to see that happen, but we’re not big enough to effectuate that. We can snap the pieces for our own benefit and for our customer, but that still isn’t doing justice to the whole world. If there was some kind of roadmap, you’d see a lot more companies buying into the open source vision, in the same manner that they’re buying into Oracle or Microsoft.”

Bachana recommends starting your company’s open source journey at the Web level. “There’s plenty of resources out there where you can get information on the different platforms available. Implement Web servers first. Hire a couple of really great developers that know the LAMP environment and task them with getting a roadmap for the back office using open source. If you start that way, you could map all your needs to readily available open source solutions out there. But it starts with getting at least one person who is a technologist that could support them in the back office.”

The goal of Red Hat High is to use free software to introduce disadvantaged kids to technology they might not otherwise be able to afford, encouraging them to pursue further education and career opportunities. Red Hat High bundles free software alternatives to expensive proprietary applications on a Fedora live CD that campers can take with them to use on any computer. The program has been in “beta” for two years, according to Greg DeKoenigsberg, Red Hat’s director of community development. “The first two years have been good,” he says. “Lots of happy kids and happy parents.” He calls it a successful trial, but one that must “scale up” in order to continue to be successful. “We’re not in the business of doing small things at Red Hat.”

Camp is in session for one week in July, on the campus of North Carolina State University. During the day, the 50 Red Hat High campers learn how to create audio and video files, design Web sites, and build 3-D animations using free software like Audacity and Blender. The evenings are reserved for field trips to bowling alleys and movies and other fun activities. DeKoenigsberg says, “We took them to the Digital Circus, a junior college level school for learning the same stuff they were learning at Red Hat High. One of the students at Digital Circus was showing them what he’s learned. ‘This is a wireframe,’ he said, and they said, ‘We learned all of that already.’ Then the professor comes in and says, ‘Let’s show you some stuff that you don’t know. Do any of you know what IK is?’ And they said, ‘Yes, that’s inverse kinematics.’” Free software gives the children the ability to learn the same techniques and skills that college-level students are learning, at a much lower cost, DeKoenigsberg says. Maya, the 3-D computer animation application that students at the Digital Circus use, can cost $7,000. Blender is free.

Now that DeKoenigsberg has two years of Red Hat High under his belt, he’s ready to take the program to the next level. “There’s some cost to holding a residential summer camp, and it doesn’t scale as well as we would like,” he says. Red Hat can provide all the financial backing necessary to translate the program into something that can serve the needs of more kids. “We would like to develop a strong affiliation with some entity that can house the kids,” DeKoenigsberg says. “We don’t run a residential summer camp as our core competency at Red Hat. But what we can do is gather a community of developers and users around open content for an important social purpose.”

DeKoenigsberg hopes that the education kids received at Red Hat High can be translated into something portable and scalable. “We putting together a deployable curriculum in a box, and making sure we have something that is turnkey.” The lessons have to be understandable and usable by the average junior high school teacher, so that more than just a handful of children each year benefit from the program. “If it’s only going to be 50 kids, it’s not worth doing,” DeKoenigsberg says. “It doesn’t make enough of an impact. The opportunity for free software to change the way kids learn is too big.” Along with the goal of making a teacher-friendly curriculum, DeKoenigsberg says it is important to get some real world feedback from actual teachers. “We are developing the curriculum in a wiki style, so that teachers can change it directly. After that, the next priority is internationalization.

DeKoenigsberg admits there are some real world challenges associated with providing kids with a free software alternative. One of those is the fact that just having a live CD is not enough for a student who doesn’t have access to a computer outside of school hours. And even for those who can use a computer at the library or elsewhere, saving files once they have been created can be a problem. At Red Hat High, each student had a scratch space on the network on which to save files. DeKoenigsberg says one of the possible future projects for the Red Hat High community is to bundle the files on a bootable, writable USB key. “The cost of that will become lower over time.”

As the senior community development manager, DeKoenigsberg is used to the idea of building a community around a cause. “People are deeply incented to spread the goodness of free software,” he says. “And the educational market is strategic to people who care about the advancement of free software. It’s strategic to the company for obvious reasons, and to the community for similar reasons.” He compares the philosophy behind Red Hat High to his company’s involvement with the One Laptop Per Child (OLPC) initiative. “We have a very calculated social consciousness. OLPC is about getting that laptop into the hands of as many kids as possible. Red Hat High is about getting the understanding of what free software can do into the brains of as many kids as possible.”

Terracotta is licensed under the Terracotta Public License (TPL), a modification of the Mozilla Public License that includes an attribution requirement. The license is not officially sanctioned by the OSI, but Terracotta doesn’t restrict access to the code, and allows modifications and redistributions as long as the code continues to be licensed under the TPL.

Pandey doesn’t want to call Terracotta a database replacement. “We try not to position ourselves [that way], because it will take us a long time to get to the point of having all the features and functionality of a database. It might be enough, though, to be a way to offload the database.” Pandey says the reason he created Terracotta was because he believes traditional databases were developed “for a world when Web apps didn’t exist. Databases were great when the access numbers were fairly limited. Fast forward 15 years and suddenly the problem has become very large scale. Peak loads can happen at strange times; databases weren’t designed to handle that complexity. What you get is Oracle coming up with solutions like rack clusters, or the customer gets fed up and says, ‘I need to write some custom software to do things like caching and offloading the database.’”

Since many businesses cannot afford the costs associated with developing custom software, Pandey says, “they’re hostage to licenses. There was an opportunity here for us, where we could come in and provide essentially what people are trying to do with that custom software.”

Terracotta was on the market for several months as a proprietary product, with less than stellar sales results. “It’s a fairly disruptive technology,” Pandey says. “We’re out there saying, don’t use a database, use us. And that’s not the first thing that comes to people’s minds. They think, ‘let me add more databases.’ We’d have to send sales reps knocking down doors to get them to go to trial with Terracotta. We were trying to change a mindset, and we thought hard: what would be a lower-friction way of doing this? What we wanted was more of a pull mode, where people get excited about the product. If you have all these expensive licensing costs up front, people are not going to try new ideas.”

When Terracotta decided to start giving away the product and the source code, things started to change. “Before we went to open source, we used to have to engage people,” Pandey says. “We don’t do that anymore. Open source is a great place for new ideas. If we find a customer that wants to do a proof-of-concept, we point them to all the resources and links. If they need any help there’s a forum where we try to maintain excellent service levels. We give them the option of calling and talking to us, but we don’t push anything on them and there’s no pressure to buy. So people are more willing to experiment.”

If Web traffic is an indicator of interest, then Terracotta’s audience has exploded. “When we launched, we would get a few hundred visitors a month; now we get 50-60,000,” Pandey says. “It’s that wonderful word of mouth, and it’s a feedback loop. We have people out there doing the marketing for us that we would have had to spend millions on to achieve the same results.”

Because Terracotta is still selling a commercially supported identical version of the product, Pandey says his biggest challenge is reining in the salespeople. “We had to address keeping a very clear wall between the sales team and the community side of the business. The temptation is, let’s call up all these people on forums and try to sell something to them. But the reason open source works is people have to feel that they do not have to buy anything to make it work. We make sure the developers have access to all the features they need — no doing any bait and switch. The other thing is we do not let the salespeople approach someone unsolicited. They have to come to us first.”

Pandey says sometimes it is frustrating to see larger companies taking advantage of the free version. “You say, wow, they have tons of money. But it’s worth the tradeoff. I would not go back to the other model.” For one thing, he says, the honest feedback from users is invaluable. “In my experience with proprietary technology, when you do a customer forum and ask for feedback, there’s always that elephant in the room. They are thinking, ‘how honest can we be, because they’re going to try to sell me something.’ So you’re dancing around. They will say, ‘I want this,’ but they’re not going to reveal the true value they get from that feature. You’re always second-guessing that. That kind of thing goes away completely with open source. The customer will tell you exactly what they need and how important it is to them, because they know it’s not something they’re going to have to pay for. You’re getting unadulterated information and you can actually build and design things that people find useful.”

And even though it was business as usual for the frontline employees, vice president Tommy Rouse knew things were very different.Rouse’s Supermarkets has been a family owned and operated grocery chain since 1959, when Tommy Rouse’s father started with one small store. The younger Rouse grew up living next door to the store, and so naturally he feels “deeply involved” in everything that happens with the business, which today has expanded to 15 stores.

Rouse’s has been utilizing ACR Retail point-of-sale (POS) systems since 1991. ACR, based in Jacksonville, Fla., has been providing software and systems integrations for grocery and drug stores since 1975.

ACR ported its ACR 5000 POS software to Linux about three years ago; before that, the company had worked with Linux for two years in a testing environment. When it came time for Rouse’s to upgrade its POS systems, ACR president and CEO John Huffman suggested thin clients and a server running ACR 5000 on Linux.

Tommy Rouse wasn’t a stranger to Linux. His IT staff had been using it for back office operations for several years, coding custom applications for data storage and retrieval. That experience, coupled with the desire to upgrade clunky Microsoft-powered boxes at each register to easily maintainable thin clients, made it easy for Rouse’s to say “yes” to Huffman’s suggestion.

In June 2003, Rouse and Huffman launched a single test store to “feel their way around it,” according to Rouse. By May 2004, they were ready to start rolling out the new system to the rest of the stores, and the switch was complete by July.

For Rouse, the top benefit Linux brings to the company is lower initial cost and lower overhead. Huffman agreed, citing the flexibility his customers have when choosing Linux as the base OS.

“[With Linux] we can supply the functionality from the server and leave the client utterly dumb,” Huffman said. The thin client “evolution” has resulted in drastically lower component costs, making the terminals so economical as to become almost “disposable.” Not only that, but stores like Rouse’s no longer need to hire highly paid technical people, since no special training is required to replace a thin client if something goes wrong. “Rouse’s keeps a couple of spare terminals in the back. If one breaks, all you have to do is plug it in — no software installation or configuration,” Huffman said.

Another benefit to using Linux is the flexibility it allows when selecting server iron. Rouse was pleased that he and his staff were able to build their own servers for less than $500 each. Because of the low cost, Rouse was able to install a separate cold backup server in most of the stores. “If we lose a server we can back that one in remotely just by making a couple of quick software changes,” he said.

Huffman is enthusiastic about the future of Linux in the POS space. The key, he said, is the ability to completely remove the operating system from the thin clients, something proprietary operating system providers do not allow, since that would cut deeply into their revenue. “Microsoft is trying to do thin clients, but they’ve got to keep their software in there,” Huffman said. “They don’t want to give it up. They live on the desktop and if that’s eliminated, they lose their market. They’re very desperate to keep some intelligence in their terminals. With Linux, we can do thin clients for effectively no cost, and Microsoft can’t. I love it.”

Jackson wanted to keep Sybase, in part because of the familiarity factor. “We looked at other databases like MySQL and SQL Server for Windows, and even considered just upgrading Sybase and running Solaris again. It made sense financially and from an ease of implementation standpoint to stay with Sybase, but with a much more current version running on Intel and Linux.” While MySQL is free, Sybase allowed Western & Southern to port its current licenses to the new infrastructure. Besides, MySQL didn’t pass Jackson’s evaluation with flying colors. “It just wasn’t good security-wise,” he says. “It’s not even close yet. It was missing a lot of the functionality that we needed.”

Even though MySQL didn’t pass muster, Linux was just fine. “We took Linux through the trials. We installed, we tested, and made sure everything matched up the way we expected it to. We had big concerns on the database side — moving the data is pretty easy but the key is the code. Anything that has database-specific code in it is a concern when migrating. By staying with Sybase, we didn’t have to rewrite the code, and that saved us hundreds of hours.”

Western & Southern reports an 80% decrease in batch cycle times on the new database servers running Linux, as well as a 60% reduction in the number of servers needed, which further reduced both hardware and software license costs.

Jackson’s biggest challenge with migrating to Linux has been version matching. “It becomes kind of sticky with all the different pieces. Versions are changing so fast now, you have to make sure vendors match the version between the database, the applications, and the operating system.”

Jackson’s experience with Linux has been so good that the company is now looking at other areas in which it could use the open source operating system. “I can’t say a whole lot about it, but we’re looking at some open source Web technologies and workflow software, and some other things.”

For other IT directors looking at migrating to an open source platform, Jackson recommends thorough testing. “They’ve got to be able to have a good test bed,” he says. “It helps to have a good understanding of how the application functions as it is now, and will it be able to bring the functionality you expect. That goes back to having people who are familiar with the application and how it will work. Especially with a database, it’s good to have people who can understand the code. Being able to play with it and experiment and get comfortable with it yourself, before you commit, is key.”

Compellent’s SAN appliance, called Storage Center, works from a GUI console that manages servers in any location using a “wizard-based” interface. Storage Center automates most of the tasks that network managers typically have perform manually, most notably the task of “tiering,” or deciding where in the SAN to store certain classes of information to create the most efficient system of data storage possible. Most of the time, data is classified according to the frequency of access, with data that is expected to be accessed less frequently stored on lower performance, less expensive drives, and more frequently accessed information stored on high performance disks. The Storage Center automates this task by placing metadata identifiers on each block of information that tell the software things like when the data was last accessed.

Guider says there was no point in starting from scratch when it came to choosing what architecture should undergird Compellent’s automated tiering capabilities. “Any smart developer knows that you want to use something already available. The operating system was the first thing we had to deal with in implementing the software.” Guider knew he wanted to use an OS with real-time capabilities to power the SAN appliance, so it could write, access, retrieve, and deliver data quickly across the network.

Guider was drawn to eCos, an open source real-time operating system, because of its highly configurable nature. He calls it an “application-specific” OS, and believes that because developers can tweak it to their apps, it delivers faster run-time performance. The fact that it is open source made it a great choice for budgetary reasons and what Guider calls “schedule savings,” because his development team didn’t have to call on a vendor to perform needed customizations.

Instead they pulled the source code, made the necessary changes to the kernel, added some driver support, “and had it up and running in a couple of weeks,” says John Veit, Compellent software engineer. Veit didn’t hit any snags with the software or with the development community surrounding eCos. Compellent had lined up distribution channels before it even began shipping its products, creating a built-in audience for the development process. “Some customers were not believing that we could have done this in the time we have done it,” Veit says. “A good part of it is due to the fact that we used the open source OS.”

Guider and Veit looked at other OS options, including VxWorks and QNX, but their colleagues recommended eCos for its stability and flexibility. “It’s not just cost,” Guider says. “I always think of engineers as a limited resource. With eCos, we used the engineers in a more productive way. This is code that has been out in the field for a long time in a variety of applications, and it is very solid. It’s a very valuable thing to have. The engineers are much more focused and we are able to have a much more robust product.”

Guider says his customer never question the fact that there’s an open source operating system underlying the SAN appliance. “We really don’t advertise what the operating system is. Sometimes people ask, but there’s never been a negative response.”

iFAX provides value-added services, support, and hardware to the HylaFAX open source facsimile server software. HylaFAX uses a client-server architecture that allows users to send documents through fax modems from any computer on the network. Nickerson says the company was “founded from the ashes” of a dotcom bust. “In 2000 I relocated from England to start this [other] company with a friend. He had something probably nobody really wanted, like many of the companies of that era. It never took off.” The company failed, but in his newly unemployed state Nickerson found much more time to work on his true passion: the HylaFAX open source project. Nickerson had been involved with the project since 1991, when it was known as FlexFax. It was his first exposure to the concept of open source software development, and he “became very enamored with the idea.” After several years of development that some said was too slow, in 1998 Nickerson was one of the community members responsible for revitalizing the HylaFAX project, when he and fellow developer Robert Colquhoun created community repository hylafax.org. After the dotcom startup died, Nickerson says he was left with a choice. “I could go back to England, or I could continue following the roots I had planted here.” He decided to make his passion into his career, launching iFAX in 2002. “That was my dream,” he says. “My goal was to build a company around this software that would allow me to pay myself and other people who had been involved: to monetize it.”

Nickerson says HylaFAX lends itself very well to a commercial business model. “We were particularly able to monetize it because HylaFAX has very clear business uses. [In the community], we were being solicited by companies all the time: ‘We’d like some support.’ But we’d never gotten organized enough. As soon as we launched iFAX, people started knocking on our door. One of our early customers was AT&T. When one of your first customers is the phone company, you know you’re doing something right.”

iFAX had such a good start that Nickerson never felt the need to solicit venture capital. “We’re completely organically self-funded,” he says. “It is our plan to remain so. Sure, it may have been a good idea to have made a bigger splash; to have come out with a larger plan earlier, but I’ve seen the dark side of VC in the dotcom era, and if that can be avoided, it should be. We already have two masters: the business itself, and the open source community. We can’t have a third.”

Nickerson considers that second master, the community, to be an indispensable part of the success of the company. Other entrepreneurs building a company on top of open source software should do the same, he recommends. “Be true to the community. Be transparent in what you do — 100%. Make it very clear to people that you won’t do anything anti-community. It’s in your best interests to make sure that the software improves and is very healthy, so don’t do anything to jeopardize that. It’s very easy to fall into the trap: let’s build some special sauce and sell that. That’s the way the proprietary software industry works, but it doesn’t map well onto an open source community. It generates hostility and they will leave you.” Instead of creating proprietary add-ons, create superior service, he says. “In our situation, we’re lucky enough to have a large line of hardware along with the software.”

The biggest challenge Nickerson faces with iFAX is one that many entrepreneurs experience: a lack of balance between work and the rest of life. “It’s especially difficult when you’re doing it in the open source software industry because there are no business hours in open source development,” Nickerson says. “They’re global. People do it whenever. It’s usual to see people all over the mailing lists on the weekends. I do it. Several of the employees do it. Nobody asks them to, but it is what we believe in — it is our passion. So it’s very difficult to maintain balance. If you talk to my girlfriend, you’ll know. The needs of the business come first, and on the evenings and weekends, the needs of the open source stuff.”

One of the principles that has guided iFAX’s rapid growth is Nickerson’s belief in “win-win” scenarios. “Open source just allows us to come in at a good price point. The fact that we can get the software exposed to businesses [in the community] and learn from them, and then contribute the improvements back to HylaFAX, is a win-win. We’re able to come in and pitch a lower cost, high performance solution, and we can do that because the software is free; there’s no line item on the quotation for software. And we come in with so many other advantages. It is open, you can modify the source, they can do it themselves. They’re not locked in. It removes a lot of barriers. It’s a great way to do business.”

Another key to success with open source is “looking to the future,” Nickerson says. “Open source project can become irrelevant very quickly, so we have diversified and taken our knowledge and built a second line of business around Asterisk, which is very similar to HylaFAX in some ways. We don’t stand still — we hit the ground running and then keep running.”

Nickerson’s number one piece of advice to entrepreneurs thinking of launching a new business based on open source software is “just do it. Try it. You just don’t know until you do, whether it’s going to take off or not.”