Someone is going to win big in Hadoop. The overall victor remains to be seen, but Wall Street’s paltry $659 million valuation of Hortonworks suggests that a Red Hat-esque approach to big data isn’t a winning hand.
That isn’t to say, however, that Cloudera’s more proprietary approach will prevail, either. Arguably the right way to consider winners and losers in the Hadoop market is by analyzing who can afford to invest most in making it simple to use.
Jumping the gun
But first, the obvious: Hortonworks never should have filed to go public. In his roadshow presentation, the Hortonworks CFO tries to color the company’s numbers as successful according to “nonstandard optics.” But in reality, the balance sheet is bleeding red, much of its revenue is tied up in only three big customers, and 43 percent of all revenue is derived from low-margin professional services, Hortonworks should have waited before going public.
Why didn’t it wait?
According to Gartner analyst Merv Adrian, “The answer may be really simple.... If not now, when? There may not be a better time.” Others argue that the decision to file early was meant to get out in front of an impending Cloudera IPO, which will generate more than $100 million in revenue in 2014. Cloudera claims to have twice the number of customers as Hortonworks and narrowing losses.
This isn’t to suggest Hortonworks is struggling. Despite seeing its valuation plummet to $659 million from a reported $1.1 billion at its Series D financing in March, Hortonworks has been growing revenue fast, generating more than 100 percent year-over-year growth.
But while Hortonworks CEO Rob Bearden previously claimed it would finish 2014 “at a strong $100 million run rate,” the reality is that it fumbled its third quarter, generating a mere $12.8 million in revenue, or roughly half of the promised $100 million run rate, as The Wall Street Journal reports.
The problem may come down to business model.
Infrastructure: Open by default
For years Hortonworks has touted its business model as its not-so-secret sauce. Like open source powerhouse Red Hat before it, Hortonworks open-sources all its software and charges customers for professional services and support subscriptions.
Hortonworks can be forgiven for believing the Red Hat model would work. After all, as Cloudera co-founder Mike Olson observes, “No dominant platform-level software infrastructure has emerged in the last 10 years in closed-source, proprietary form.”
If you want to play in the infrastructure space, you have to be open source -- period. Hadoop is the heart of data infrastructure. Of course it has to be open source.
There are a number of reasons for this insistence on open source, but the rising importance of developers offers perhaps the most cogent explanation. Developers expect open source. While a business user may be content to pay for her apps, the developer wants her software to be free, available under flexible licensing terms.
Open is not enough
The problem, however, is that “open” can’t be the complete answer, not if we want the software developers to remain in business long enough to write the software.
As Olson points out:
It's pretty hard to build a successful, stand-alone open source company. Notably, no support- or services-only business model has ever made the cut. Red Hat, the apparent exception, isn't: The company rode its closed-source Red Hat Network offering to dominance, effectively crushing the competition before releasing that hosted infrastructure as open source in 2008.
As such, Cloudera has tried to strike a balance between proprietary software and open source. Having walked this same tightrope for years, I know how difficult it can be. But I would also agree with Olson that it’s essential.
At least, it's essential for the businessperson. But what about those developers? I asked Doug Cutting, Hadoop’s founder (and Cloudera executive), what he thought about this blending of proprietary and open source code. His answer is illuminating, as it suggests that both Hortonworks and Cloudera may be right, albeit at different times:
Over time enterprises will realize that a high-quality support offering has unique value and is worth a premium. In the pre-open-source-software world this was clouded by inflated software licensing fees that subsidized support costs. As licensing fees now shrink, enterprises still need just as much vendor attention, including bugfixes and development of new features.
Payment for that needs to come from somewhere, and it will gradually shift from software licensing fees to support subscriptions. But that change won't happen overnight and may never be total, so in the interim, it may be wise to adopt a hybrid strategy like Cloudera's.
Halfway open source
So far, the market seems to favor Cloudera’s model. Cloudera raised its last gargantuan round of financing at a valuation of more than $4 billion. But it’s not yet clear that these billion-dollar unicorns will fly, especially if fueled by the promise of customers paying for free software.
While Hortonworks points to its 100 percent open source policy as a market strength, Wall Street’s $659 million valuation begs to differ. In part this is because Hortonworks may be putting too much faith in the “Expand and Renew Phase” of its business model (screenshot taken from its roadshow presentation), the very phase that 451 Research analyst Matthew Aslett called an “open source dilemma” years ago.
The cause of this dilemma? Early subscription revenue is highly unprofitable because the early years of the subscription are when a customer needs support the most. But eventually they grow comfortable with the technology and stop renewing ... exactly as the customer becomes profitable.
This led Olson to conclude, “You can no longer win with a closed-source platform, and you can't build a successful stand-alone company purely on open source.” Perhaps Hortonworks will eventually reach the same point.