If you read the Red Hat website, you’ll find pages describing their attitude toward open source, collaboration, and more. It reads pretty much like every other marketing spiel from every company online today. There’s something different about Red Hat, though: they actually believe this stuff. Not only do they believe it, they live it every day.
I spoke to Red Hat CEO Jim Whitehurst recently about the open source culture at Red Hat and he told me it is a journey, not a destination. According to Whitehurst, the tenets of open source permeate all aspects of the culture at Red Hat.
Whitehurst opened our conversation by stating that the last eighteen months have been a tipping point for Red Hat. According to him, they’re “no longer fighting an uphill battle for credibility.” Nowadays the conversations he’s hearing with customers focus on the issue of price versus performance, rather than whether Red Hat is a viable player in the enterprise marketplace.
Red Hat’s big news is that they’ve broken a billion dollars in revenue, and are arguably the first all-open source company to do so. I’ve read some disagreement with the notion that Red Hat is a “pure play” open source company. I asked Whitehurst whether their earnings claims were just a matter of semantics: is Red Hat really a billion dollar open source software company, or a billion dollar support and services company? “We sell software,” was his immediate response.
Rather than a specific, static piece of software, Red Hat sells you a subscription to their software. While the Red Hat source code is freely available, the compiled bits included in your subscription are not free. What you buy from Red Hat, says Whitehurst, is software that has been rigorously tested, known to be good, and is fully supported.
According to him, the long-term support of Red Hat’s enterprise software represents 6% of the company’s overall costs. Red Hat employs “an army of engineers,” says Whitehurst, “maintaining 10 year compatibility.” The company spends even more money on engineering new solutions every year.
I next asked Whitehurst “How does the open source culture affect Red Hat internally as a corporation?” Specifically, I wanted to know how have “the tenets of open source” affected management processes? This was, in many respects, the most interesting part of my conversation with him.
“The traditional hierarchical org structure was developed to control, and steward fixed assets,” Whitehurst told me. The traditional hierarchies do not tend to innovate. “When information is your primary product, hierarchy isn’t the best way to drive.”
To illustrate how the open source culture influences everything at Red Hat, Whitehurst told me a story about an early experience as the company’s CEO. When he started, the company had — to his mind — a rather weak mission statement. Coming from a traditional management background, Whitehurst’s first reaction was to do what CEOs all around the world do: gather the executive team, have an off-site retreat with a paid facilitator, and develop a new mission statement. This new statement would then be broadcast to the employees.
“Hold on, sparky,” was effectively what his executive team told him. “You need internal buy-in,” he was told. An internal collaboration site was established — mailing list, wiki, blog — to discuss the mission statement with staffers who were passionate about that issue. Once the infrastructure was put in place, the discussion was entirely self-organized, and about 15 to 20 people really dug in. “Iterate, iterate, iterate” is how Whitehurst described the process.
What was particularly interesting to him was that this ground-up process saw participation from all parts of the company: not just developers writing code but also artists and designers, UI and UX experts, and more. The end result was a mission statement that had tremendous buy-in throughout the organization, and one that didn’t have to be “sold” by management.
“It takes us so much longer to make decisions because so many people are involved,” Whitehurst said. “Once a decision is made, though, we have no problems with execution.”
He told me that it took him about a year to really understand “the open source way,” and another year to realize that as a leader he needed to be an internal catalyst. “Open source is not about democracy at all. It’s a meritocracy,” he said. As such, his employees expect two things from him:
- solicit feedback before decisions as much as possible
- assume accountability for decisions he drives
Whitehurst claims that this forces much more robust conversations, but leads to significantly better results.
Another example of the meritocracy of open source at work at Red Hat: Whitehurst has never sent a memo. Every communication needs to support response from recipients, rather than be seen as a pronouncement from on high. At Red Hat, this takes place on one of several internal mailing lists. They’re not afraid to set up new microsites (blogs/wikis/lists) for focused participation on specific issues, either.
I’ve been a participant in open source communities for more than a decade, so much of what Whitehurst shared was common sense to me. But it’s clearly not business as usual at the executive level of a billion dollar company. I asked him “Do you only hire open source “true believers” at the management level? Is there strife between “true believers” and traditional business school grads?”
For a long time they tried to groom the people they hired, Whitehurst told me. He shared the anecdote of one employee who quit after their first day: “This is chaos, I can’t do it,” that person said. Now more than 50% of all hires come from internal referrals. This includes all levels of the organization, from entry level through the executive level. According to Whitehurst, this helps ensure that new employees better fit the culture, and results in a lower attrition rate than seen at other tech companies. Interestingly, in the last several years only one person at the VP level has left the company, Whitehurst told me.
Following up on Whitehurst’s opening remarks, I asked if Red Hat’s continued growth has affected their sales strategies at all? Specifically, are new opportunities opening now that Red Hat is a billion dollar company? The short answer is “not really,” but the longer answer Whitehurst provided was that over the last couple of years Red Hat’s success has made people feel more comfortable that they’re a long-term player. This has led to a self-reinforcing cycle, the results of which are clear.
Linux is displacing older proprietary UNIX systems (HP-UX, AIX, Solaris) at an incredible pace. I’ve often been curious how long it will be until the rest of open source catches up? For example, who’s really threatening Oracle for traditional relational databases? Whitehurst’s answer here wasn’t quite as direct as I had hoped. He highlighted the “explosion of open source contributions from Web 2.0 companies like Google, Facebook and Twitter” as examples of people looking to solve their problems in “open source ways”.
“Open source is ultimately where all companies are heading,” Whitehurst stated flatly.
Companies are “going to solve problems in new ways,” rather than just commoditizing things that already exist like relational databases. The bleeding edge of innovation is driving entirely new ways to solve problems. With respect to my example about Oracle, Whitehurst pointed out the explosion of NoSQL databases and “Big Data” computing needs. It may take some time for these new solutions to trickle down into the commodity computing space, but it’ll happen eventually.
My conversation with Whitehurst was thought provoking, and covered so much ground that I found myself out of time long before I was out of questions. Whitehurst indulged me with a few follow-up questions by email, which I share here in their entirety.
Scott Merrill: Open source is open to anyone, but we still see it as a primarily male, Western approach. Is Red Hat making any effort to remedy that?
Jim Whitehurst: I don’t see open source as being more suited or relevant to a specific region, country or ethnic group. That said, open source communities are generally built from use (i.e. a % of users become contributors), through social or relational networks, exposure from college or a desire to contribute to the greater good. As many projects start as a bunch of friends or an individual with a desire to solve a problem for everyone, projects in their early stages tend to be less diverse. As projects grow, get exposure and use they tend to diversify. An example that did not follow this trend is the launch of oVirt.org – an open source alternate to VMware. That project has members active from the US, Israel, China, Japan, India, etc. Linux is another diverse example that has contributions from just about every country and ethnic group in the world.
SM: Now that Amazon has blessed Eucalyptus, do you feel any pressure to divert resources away from OpenStack?
JW: We actually announced last week that we have joined the OpenStack Foundation as a Platinum Member and will be continuing to invest resources into that community. In some research on contributions made to OpenStack recently, Red Hat ranked third among all corporate contributors and we continue to plan to invest here as we see OpenStack as complementary to our cloud portfolio and approach.
SM: How much of Red Hat’s internal IT runs on proprietary software?
JW: Perhaps a better question is how much of our internal IT is open source-based. We use Red Hat Enterprise Linux almost exclusively as our operating system platform. 100% of our development and 60% of our production infrastructure is virtualized, almost exclusively on Red Hat Enterprise Virtualization. Our Enterprise Service Bus, application tier, and our redhat.com web site are built with JBoss. We have over 4,000 associate laptops running Linux, with well over half that number running Red Hat Enterprise Linux and the balance running Fedora. Our email and calendaring system, chat system, Intranet, office suite, browser, and email clients are open source. We also use open source solutions for a variety of speciality applications, such as document collaboration and vacation tracking. We recently deployed an open source phone solution for our remote associates and plan to offer it to the entire organization. In short, we use our own products and we work hard to be a beacon for the implementation of open source solutions.
SM: Does Linux on the desktop make sense any more? If so, what is Red Hat doing to make it happen?
JW: Red Hat does not provide client/desktop products for the consumer market and does not plan to pursue this strategy in the foreseeable future. We do, however, today offer the Red Hat Enterprise Linux Desktop and Fedora, which are successfully meeting the desktop needs of our chosen markets and customer base.
SM: Where do you want to take Red Hat next?
JW: So far at Red Hat, my focus as been squarely on our datacenter business, trying to execute really well on our core offerings. I think the future is in the cloud and big data. Our acquisition of Gluster last year and the work we have done around these storage offerings gives us a huge opportunity — an opportunity that I think could actually be bigger than the Linux business.
Red Hat has come an awfully long way since the company started in 1993. They’ve had their fair share of missteps, to be sure, but on the whole they’ve been remarkably successful. Their broad participation in many upstream open source projects demonstrably benefits Red Hat’s bottom line, but it also improves the state of the art for everyone. Whitehurst’s observations about open source culture and its effects on corporate culture should be instructive to everyone.