Engine Yard: The Ruby on Rails Company Salesforce Didn’t Buy

1/31/11Follow @wroush

Salesforce.com startled Silicon Valley techies last month by acquiring Heroku, a San Francisco startup born just three years earlier at Y Combinator, the venture incubator program. Heroku’s specialty is hosting Web-based applications written in Ruby, a programming language so powerful yet easy to use that it has become the tool of choice for developers building consumer-facing Web services. Heroku had won a rabid following among fellow Y Combinator companies and other startups, perhaps helping to explain how its founders—who had raised just $13 million in venture capital—persuaded Salesforce.com to fork over $212 million for the 30-employee property.

But just a few blocks away from Heroku in San Francisco’s South of Market neighborhood is another Ruby hosting company that’s been around longer than Heroku and has a lot more engineers, support staff, and paying customers. It’s called Engine Yard, and with almost $40 million in venture backing from Benchmark Capital, New Enterprise Associates, Amazon, and other firms, it’s assembled a user base of more than 1,800 companies for its “platform as a service.” Its customer roster includes names like Bleacher Report, BuyWithMe, Get Satisfaction, MTV, Oneforty, and Path.

So why did Salesforce.com go after upstart Heroku rather than established Engine Yard?

The answer may have as much to do with Silicon Valley politics as anything else. I got part of the background from CEO John Dillon and chief technology officer Tom Mornini when I visited Engine Yard in mid-December. Dillon, I knew, has a history with Salesforce.com. He was its president and CEO from December 1999 to November 2001, when he was ousted by founder Marc Benioff. Salesforce.com said at the time that Dillon departed by mutual agreement, but the real story, Dillon told the San Francisco Chronicle in 2002, was that “Marc decided that he wanted a turn at the wheel.”

Whatever the case, there’s no love lost between Dillon and Benioff. Dillon—who has what you might call an unguarded speaking style—has gone on the record as saying that Salesforce.com’s own Force.com cloud application platform is “crap” that “no self-respecting developer would use.” So once Salesforce.com decided it wanted to get into the Ruby hosting game, Heroku was likely the more palatable choice, Dillon told me. “There are only two companies you can buy to get this, and one is Engine Yard,” he says. “I don’t think Marc wanted to write a big check to me.”

Of course, there’s a lot more to Engine Yard’s story than its rivalries with Heroku and Salesforce.com. The rise of Ruby and the closely related tool Ruby on Rails, a package of open-source software that greatly simplifies the creation of database-driven websites, has transformed the way Internet entrepreneurs implement their visions. Ruby on Rails was designed specifically to allow rapid product iteration—the so-called “agile” development strategy—with the result that developers can release and test several versions of a Web-based service in a single week, sometimes even a single day. “Ruby on Rails is somewhere between 2 and 5 times more efficient for coding a website than anything previously—and we believe it’s closer to 5 times,” Mornini says.

Engine Yard has been contributing to this transformation since its beginning: Mornini and co-founder Lance Walley set up the company in early 2006, less than a year after Danish programmer David Heinemeier Hansson of Chicago-based 37signals released Ruby on Rails. It’s a classic story of programmer-entrepreneurs seeing a burgeoning gold rush and deciding to sell picks and shovels—but in a way that’s created a lot of actual value for its customers.

“Often, tech companies are started by a couple of Stanford MBA guys saying ‘We should be rich, let’s write a business plan and go down to Sand Hill Road and get some VCs and boomity-boom, we’ll sell out and we deserve it,’” says Dillon, who became Engine Yard’s CEO in January 2009. “The problem is that nobody knows whether there is any value in that, or whether the dogs will even eat the dog food. Tom and Lance built a company the old-fashioned way. They encountered demand and they worked their tails off to meet it.”

In 2005 Mornini and Walley were partners a Sacramento, CA-based software consulting company called Quality Humans, teaching big companies about software version control and other practices that would bring their systems into compliance with Sarbanes-Oxley regulations. One day a copy of Hansson’s 2005 book Agile Web Development with Rails showed up in Mornini’s mail. (He’d ordered a pre-release version from Amazon and then forgotten about it.) “I started reading through it and I was just blown away,” Mornini says. “The root philosophy they had is that there isn’t more than one way to do everything; they said ‘You should do it our way,’ which I saw as an unbelievably powerful mechanism for keeping programmers aligned and being able to work on multiple projects at once with one code base.”

At the same time, Mornini says, Ruby on Rails was a very clean language. “I saw these beautiful examples where you literally didn’t need documentation, because you could just read the code and it was entirely obvious what different sections were doing,” he says. Mornini was so excited that he used a redeye flight from California to Washington, D.C., where he and Walley were involved in a federal consulting gig, to try reimplementing the website for one of his failed previous startups using Ruby on Rails. “I got off this five-hour flight and had about 80 percent of the site rewritten,” he says. “I was amped and buzzing. I showed up at the apartment we were renting, and Lance was like ‘What the hell is going on with you,’ and I said, ‘Check this out, we need to get into the Ruby on Rails business today.’”

Walley and Mornini started consulting on Ruby on Rails development projects, and “the business started flying in,” Mornini says. That was when the pair realized that established Web hosting services like Rackspace weren’t equipped to help developers get their Ruby on Rails applications out to the world. “None of the traditional hosting services knew a darn thing about it,” Mornini says. “We decided we had the opportunity to be the commercial entity in Ruby on Rails.”

So Quality Humans morphed into Engine Yard, and Mornini and Walley started building a data center—what would now be called a computing cloud—and hosting other companies’ Rails applications. “It was unbelievable how it took off,” Mornini says. “Three months after we started, Benchmark Capital contacted us and said, ‘We’ve noticed that all the interesting companies coming to us now are using Ruby on Rails and it seems like almost every one of them has chosen to host with you. We want to know who you are.’ That was a meeting we didn’t turn down.”

By December 2007, Engine Yard had moved to San Francisco and collected a $3.5 million Series A round from Benchmark. That was followed by a $15 million follow-on investment just six months later—with Amazon and New Enterprise Associates now joining in—and a $19 million C round in October 2009, which brought in additional investors DAG Ventures, Presidio Ventures, and Bay Partners. At the same time, other San Francisco-based cloud startups such as Heroku and Joyent were also winning venture funding, reflecting the general enthusiasm in Silicon Valley about the new innovation the cloud-services model was enabling.

But while Joyent and other cloud providers were more language-agnostic, Heroku and Engine Yard stayed focused on Ruby. Mornini says the power of Ruby on Rails—as compared to Java, .NET, C, Python, and other application languages and frameworks—is that it’s paired with a development methodology that emphasizes pleasing customers over planning. “You don’t stress about the future,” says Mornini. “You work on the next feature only, and you rely on customer feedback at each stage to select the next important thing.”

Ruby on Rails developers have time to focus on customers in part because the framework makes it easy to shift certain burdens to providers like Engine Yard. “You want developers to be like best-selling authors—you want them writing The Da Vinci Code and Harry Potter, not setting up variable definitions and all the things programmers used to have to do,” says Dillon. “Now, if you have a browser, you have access to a data center. We go out and source all of this componentry that is necessary to deploy and manage these applications. We optimize and harden it and serve it up as a service, so that you only pay for what you use. It’s cheap until you become very successful, and at that point you’re happy to pay the money, because the development team can concentrate on writing this best-selling novel.”

Engine Yard’s own story has taken a couple of twists and turns. As the economic crisis cut into demand, the company retrenched, laying off 15 percent of its employees in early 2009. That was also when Dillon came in, replacing Walley as CEO. “We needed someone with the experience to grow the company way beyond my/our abilities,” Walley blogged at the time. He has since left to head Needham, MA-based Chargify, which manages recurring billing for Web companies that depend on credit-card-based subscriptions.

Meanwhile, the company decided that it didn’t make sense to own and operate the two data centers—one on each coast—that made up its private cloud. It closed the facilities and outsourced its entire infrastructure to Amazon Web Services and Terremark, a large enterprise hosting provider. Mornini says the move allowed the company to concentrate on maintaining the stack of technologies, such as application and database servers, needed to support Ruby on Rails applications. “We may be the first large-scale hosting company ever to disband and outsource,” Mornini says. Engine Yard makes money now by adding a small premium to Amazon’s or Terremark’s pay-as-you-go prices. “If Amazon is 10 cents per hour per virtual machine, we are maybe 12.”

Demand for Ruby support has bounced back as startups using the agile methodology to build Web-based services have multiplied—which may explain why Salesforce.com wanted in. Mornini says he sees the Heroku purchase as a positive sign for the entire Ruby on Rails community. About a month before Heroku got scooped up, he points out, Palo Alto, CA-based VMware quietly announced that its new “Open PaaS” platform would include support for Ruby on Rails. “To me, that was the first brick in the wall, and now we are delighted that we’ve got two,” Mornini says.

“For us [the Heroku purchase] is great because it’s validation,” Dillon adds. “Now that we have people like Salesforce and VMware as industrial-strength sponsors, it’s wind behind our backs for Ruby on Rails.” And Dillon insists he isn’t worried about stronger competition from Heroku now that it has Salesforce.com’s resources behind it—in fact, just the opposite. “I think it’s going to get mucked up and lost and compromised inside a behemoth company, and that will work to our advantage.”

Wade Roush is a contributing editor at Xconomy. Follow @wroush

By posting a comment, you agree to our terms and conditions.

  • Lonny Eachus

    Interesting how this whole story about the founding and development of Engine Yard has edited out the very existence of Ezra Zymunctowicz, who was a co-founder and a very essential part of Engine Yard’s development.

    I know the egos of company heads can tend to distort things a little, but this is completely ridiculous. No mention of Ezra at all?

    Then how can we credit any of the rest of the story?

  • http://engineyard.com Jayson Vantuyl

    As one of the founders of Engine Yard who was also not credited, the omission doesn’t particularly bothersome to me. I don’t think the purpose of this article was really to edit anybody out. Sure, Ez and I deserve credit, but that’s not the meat of this article.

    That said, I also have another theory. I don’t think that Salesforce bought Heroku because they just thought Ruby is cool. I think they bought them because they want to provide a solid platform to provide apps that integrate with Salesforce. They want to be a one-stop-shop for their customers to host Salesforce apps. This is reasonable and good for all involved.

    That said, Engine Yard was not so attractive for that. Heroku is trying very hard to be an infrastructure platform. They don’t really have people to help integrate you with the Cloud, architect your app, or make any number of hard decisions that EY helps its customers make. They’re just there to host.

    From Salesforce’s point of view, EY likely has tons of people that would have to be tossed overboard. There would have been tons of friction because we’re about building Ruby, the Ruby community, and businesses that use it.

    If my supposition is correct, Salesforce needs a solid hosting platform that just happens to be built around Ruby. From that perspective, it makes sense to buy the smaller company with less people to fire, rather than the more established company with a strong identity that would have to be dismantled.

    Had Ezra’s work at VMware been ready when they needed it, they might not have even bought Heroku. I really think this is just about their needs at the time and finding the best option that fit those needs.

  • http://engineyard.com Tom Mornini

    Thanks for your article, Wade!

    Lonny, I don’t believe that any “editing” or “ego” are at play here. John Dillon and I were interviewed, and Wade wrote the article.

    No effort was made to bury Ezra’s contributions.

    I wish that every employee of Engine Yard was mentioned in the article! Engine Yard’s success reflects the work of our entire team. Our founders and executives are less important than the amazing team we’ve built.

  • http://www.xconomy.com/author/wroush/ Wade Roush

    @Lonny, @Jayson: In preparation for this piece I had a long meeting with Tom Mornini and John Dillon, and what you read here is what they related to me. I’m sure they didn’t mean to leave anyone out. I had many subjects I wanted to cover in the interview, so naturally they could only provide a thumbnail version of Engine Yard’s history. I do regret not having time to conduct additional research into the company’s founding.

    @Jayson: I think you’re right that it would be have been far more difficult for Salesforce to successfully integrate Engine Yard, for both cultural and engineering reasons. Heroku is already built around the multi-tenancy model Salesforce uses, for one thing.

  • Lonny Eachus

    “Edited” was the wrong word, and I should not have used it. But as a reader, I got the impression that the article was trying to give a pretty complete description of how Engine Yard was born and grew, when in fact some very big parts were left out. It did leave me with a bad impression.

    I just think the article could have done a better job explaining that its “history” of Engine Yard is anything but complete.

  • Think Again

    Whatever the politics, in my opinion as a Rails developer, Heroku is better because it lets people start developing on it for free. Not so with Engine Yard.

    Now that Heroku is part of Salesforce, we can look forward to a slow death to Heroku’s coolness. Smartest move for Dillon would be to offer a free starter package for developers on Engine yard. Will get people flocking to it in troves.