Part 2 of a two-part series on the challenges of making money from open-source software. Part 1 ran earlier this week.

When Oracle Corp. acquired Sun Microsystems Inc. in 2010, it picked up a bit of a hot potato: an open-source database. The events that followed illustrate just how difficult it has been for companies based on open source software — arguably the biggest software trend of the last decade — to find a sustainable business.

Two years earlier, Sun had acquired MySQL, the open-source relational database engine ranked as second only to Oracle’s as the world’s most popular database. MySQL was a potential rival to Oracle’s cash cow, and its developers feared that the database giant would starve it to death. So they took out an insurance policy, building a drop-in replacement called MariaDB that quickly flourished, capturing customers such as Google Inc. MariaDB Corp. was founded to commercialize its namesake product with a business model built on packaging, support and training.

But last summer the company jolted some of its supporters by announcing a release of one of its enterprise-oriented products under a new license that gives software developers a two- to three-year window to monetize their creations before it reverts back to open source — outraging some purists. Michael Howard, the veteran Silicon Valley executive who became MariaDB’s chief executive early last year, said the company is still committed to open source. But it’s also committed to its investors, who have bet $48 million on it so far. (And on Monday, they added $27 million more.)

That dilemma is common to companies that try to pursue a business model upon community-developed open-source software: Customers won’t pay enough for support and training services alone to sustain a growing company, but imposing proprietary licensing features on an open-source project risks alienating the developers who are so critical to its success.

The result: It’s less clear than ever that open source will actually end up producing successful companies. “If Red Hat tried to raise money on Sand Hill Road today, no one would fund them,” said Joe McCann, CEO of Nodesource Inc., which has raised more than $15 million to build tools for the JavaScript programming language.

Will there be another Red Hat?

That doesn’t mean entrepreneurs have given up trying, because the potential is so tantalizing. Red Hat Inc. has built a $15 billion market capitalization based on a product line that’s entirely community-developed. Hortonworks Inc. continues to provide the strongest evidence yet that the Red Hat model is transferable. On Thursday it reported first-quarter revenue growth of 35 percent and said it’s on track to hit $240 million in revenue this year.

Hortonworks’ business model is, in many ways, strikingly similar to Red Hat’s. Both companies specialize in simplifying and integrating complex ecosystems. Both pick and choose which platforms and projects to support so customers don’t have to. Both offer customers peace of mind by guaranteeing that their product suites will be stable and interoperable. “Our model is packaging innovation downstream,” said Shaun Connolly, Hortonworks’ chief strategy officer. “We’re not in the business of supporting all Apache products. We spend time picking the right versions, certifying them and delivering multiyear maintenance. Our customers and partners look to us for a coherent roadmap.”

Although Hortonworks doesn’t develop any proprietary software, it invests comparable resources in open-source projects, and its income statement looks more like that of a software company than a service provider. Three-quarters of Hortonwork’s first-quarter 2017 revenue came from subscriptions, a number that was similar to rival Cloudera Inc.’s. Hortonworks’ 71 percent first-quarter gross profit margin was almost identical to Cloudera’s and not far behind Oracle’s 78 percent. The company is on track to crack the $200 million annual revenue ceiling faster than Salesforce.com Inc., Cloudera, Red Hat or Oracle.

Pivotal Software Inc., Alfresco Software Inc. and Automattic Inc. are among the other well-funded companies trying to make a go of the pure-open-source business. But none has yet had to endure the scrutiny of public markets, and skepticism is growing that such businesses can succeed outside of niches or complex platform ecosystems like Linux or Hadoop.

As a result, some companies that have tried the pure-open-source approach are now retrenching as they look for new business models. With open-source code undergirding most of the largest web-scale companies and penetrating deeper into the enterprise, open-source-based business are increasingly pursuing a hybrid strategy of a community-code foundation with proprietary extensions that command a license or subscription cost.

“We put a lot in our free edition, but we were continually competing against that product, and were challenged to express to customers the value of a commercial edition” that was functionally identical, said Clint Oram, co-founder and chief marketing officer of SugarCRM Inc.

Founded in 2004, SugarCRM spent its first 10 years selling services around an open-source version of its customer relationship management suite, but three years ago it made the decision to stop enhancing the open-source version and move forward with future editions under a commercial license. The company continues to issue patches for the community edition, but its development efforts are entirely commercial.

Echoing other open-source veterans who were interviewed, Oram said business models based upon pure open-source software can work as long as the market doesn’t get too big. “Based on my experience, a business approach in which revenue is based entirely on support or maintenance scales up to about $50 million and then starts working against you,” he said. “The place you can differentiate is in products and features.” The fact that sales executives – which are SugarCRM’s core market – didn’t much care about open-source development complicated things further, he said.

MariaDB’s Howard was even more emphatic. “To meet the demands of the enterprise you can’t have a pure open-source model,” he said. On the contrary, Howard added, enterprises expect to see value-added features from their core infrastructure vendors and care little about who owns them. “Open-source companies have to find a way to embrace that challenge, and it’s not going to be based on near-zero investment in development and just relying on service contracts,” he said. “I’m deliberately making MariaDB a product company and not a service company.”

Hortonworks’ Connolly agreed that pure-play open-source businesses are hard to build outside of complex, rapidly evolving platform markets. “If it isn’t a platform play, 100 percent open-source isn’t necessarily right,” he said.

Tide has turned

Despite the business challenges, executives agree that open-source platforms will increasingly dominate infrastructure markets. Customers are demanding it. “When I meet with some of the biggest companies in the world, I see that they care deeply about open source,” Howard said. “They also care very much about the stability of the companies that are associated with that open-source product.”

One of the advantages of open-source development is rapid product evolution. For example, the Hadoop community has released five new versions in just the past nine months. That compares with the one- to two-year release cycles of commercial software for a single update. The pace of innovation in Hadoop is so fast, in fact, that Hortonworks last year said it would throttle deliveries to customers in order not to overwhelm them.

Community development can also yield better products. “Instead of having features drummed up in [Oracle Founder] Larry Ellison’s head, they’re a reflection of the community,” Howard said.

Then there’s the viral propagation effect. Software released under a try-before-you-buy “freemium” model percolates far more quickly through the market than software sold to each enterprise individually.

“The time to get a product out into the market has effectively gone to zero because open source gets it in front of customers almost immediately,” McCann said. Once organizations adopt the freemium version, “they discover they need better auditing, security, monitoring and the like,” he said. “But they’ve already adopted the software.”

That’s what SugarCRM is betting. “What works best is when you have an ancillary product that drives its own value but is separate from what your customers pay for,” Oram said. Differentiating based upon proprietary extensions also has the benefit of not subjecting every decision to “a tug of war with your community over products and features,” he said.

So far, open-source-based businesses haven’t proven to be as lucrative on the bottom line as the vertically integrated models that spawned Oracle and Microsoft, but the market is still young. Total market sizes are potentially bigger, but margins are tighter. Snowflake Computing Inc. CEO Bub Muglia recently observed that Red Hat may be successful, but after 24 years it’s one-tenth as profitable as Microsoft. Muglia would know. He worked at Microsoft for 23 years.

One new avenue investors and entrepreneurs are now exploring is open-source-as-a-service. “Cloud opens up a whole new opportunity to monetize,” Hortonworks CEO Rob Bearden said during the company’s earnings call. On the other hand, cloud infrastructure vendors are poised to steal that market by delivering open-source products without contributing to them, while skirting licensing restrictions with resource-based pricing.

And perhaps the best models haven’t even been tried yet. Forrester Research Inc. Senior Analyst Milan Hanson noted that two of the biggest developers of open source software – Facebook and Google – aren’t even software companies. “Might open-source-based businesses similarly find creative revenue sources besides the obvious services revenue?” he asked. A lot of companies and their investors sure hope so.

Photo: skeeze/Pixabay