Hadoop is booming -- its vendors, not so much

Hortonworks had a better quarter than generally acknowledged, but can the open source company sustain itself over the long haul?

fireworks boom explosion explode celebration
Michele Ursino/Flickr

Forrester Research says enterprise adoption of Hadoop is becoming "mandatory." But Hortonworks’ first earnings release as a public company suggests heavy lifting lies ahead. After missing earnings by 20 cents and registering lower revenue growth than analysts expected, Hortonworks has a ways to go before it can proclaim itself a big data darling.

Or does it?

While many analysts looked askance at Hortonworks’ numbers, others -- including Wells Fargo analyst Jason Maynard -- see mostly silver linings. The reality of the Hortonworks release is both positive and negative, and it's far more indicative of the difficulty of building an open source business than it is of any particular Hortonworks failings.

[ Reap the benefits of Hadoop with InfoWorld's Hadoop Deep Dive. ]

First, the numbers

For better and for worse, Hadoop has long been the poster child for big data. So when Gartner finds that 73 percent of enterprises have near-term plans to invest in big data (or already have), the reality is that many of these companies mean “Hadoop” when they say “big data.”

Hortonworks, one of the top two Hadoop vendors, stands to benefit from this market shift. Seemingly, it already is.

While many (including myself) called out Hortonworks for missing earnings, the reality, according to Maynard, is very different. Calling out billings growth of 148 percent year over year, and revenue (non-GAAP) nearly doubling, Maynard writes in a research note:

We continue to believe that demand for big data technologies will expand, and Hadoop will become mainstream. We think that Hortonworks is well positioned to be a leader in this burgeoning market. The outperformance in the quarter reinforces our view that Hortonworks has a strong competitive position supported by its open source business model, large number of Hadoop committers, and strong partner network.

The problem with measuring Hortonworks’ performance is that it’s not a software business, though analysts want to treat it as such.

One of the strongest indicators of Hortonworks’ business is not only new customer growth, which was impressive for the quarter (99 new subscription customers), but also the rate at which existing customers reinvest in the relationship. By this metric, Hortonworks is doing really well, as customers upped their financial commitments to Hortonworks by 144 percent on average over the last year.

That’s the good news.

The bad news, however, is that the company is losing more money even as it makes more money. The other bad news is that Cloudera, the top Hadoop vendor, is doing twice the revenue that Hortonworks is. Also, Cloudera insists its revenue is GAAP-approved, rather rather the non-GAAP, which is how Hortonworks prefers to report its earnings.

While some Hortonworks initiatives -- the Open Data Platform comes to mind, rightly excoriated by Gartner analysts Merv Adrian and Nick Heudecker -- are an exercise in vanity and futility, others such as the Data Governance Initiative show much more promise and leadership.

As Ovum analyst Tony Baer writes of DGI, Hortonworks has taken on the “thankless task” of corralling industry participants to drive data governance activities that are “likely to spawn future Apache projects that, if successful, will draw critical mass participation for technologies that will be optimized for the distinct environment of Hadoop.”

It’s an example of Hortonworks at its open source best and hopefully a sign of good things to come.

It does not, however, ensure eventual riches.

Code is currency

In general, she who contributes most controls and influences most in open source. Instead of intellectual property, contributions are the currency that generally earns the biggest returns, as Red Hat has shown in Linux, OpenStack, and other communities to which it contributes. As Hortonworks CEO Herb Cunitz declares, “Unless you have influence [with a project through contributions], it's not a monetizable business.”

It may not be a “monetizable business,” anyway.

There’s a reason that only Red Hat has managed to earn $1 billion a year in GAAP revenue: It’s hard to sell “free.” Or as former Kaplan CTO Jon Williams once outlined:

The [happier] he is with his commercial open source software, the less likely he will be to pay for it. Why? Because his developers will acquire the expertise over time to support themselves and because the product will mature to the point that support will be less necessary.

Hortonworks touts it as an advantage that it doesn’t sully its services with proprietary software; to be clear, this resonates with some buyers. But after 15 years of selling open source subscriptions at a variety of companies in different areas of the software stack (operating system, database, applications), it’s very clear to me that the service-and-support model won’t sustain future Red Hats.

Former XenSource CEO Peter Levine persuasively argues why there will never be another Red Hat, highlighting along the way Red Hat’s relative underperformance vis-a-vis more proprietarily inclined peers. As he concludes, “the business model simply does not enable adequate funding of ongoing investments.”

Coming back to the code-as-currency argument, this means that Hortonworks, prolific as it has been with venture capitalists’ money in fueling impressive investments in YARN and other Hadoop staples, won’t be able to sustain these investments on its own billings growth, hefty as that has been.

Right now, Cloudera gets paid more than Hortonworks for the code Hortonworks contributes. That’s not a long-term winning strategy.

Give it time?

Of course, Hortonworks could defy open source gravity and become the next billion-dollar open source company (in terms of GAAP revenue, not valuation). To get there, it’ll need to stop pretending it doesn’t have to play by the same economic rules as other public companies.

While some analysts argue, with the company, “The Hortonworks model is complex at the moment and I don't believe it should be evaluated based on the normal expectations of public companies,” the reality is that Wall Street has little patience for such deviations from the norm. Red Hat spent years trying to get ahead of its revenue and finally has some leeway to acquire companies that aren’t accretive.

Red Hat, in short, can finally afford to invest strategically, not merely tactically.

Hortonworks is a long, long way from that point. It may get there, but it has chosen a model that gives it marketing bragging rights but even stronger financial headwinds. Wikibon analyst Jeff Kelly, speaking of another corner of big data, NoSQL databases, offers insight that applies equally to Hortonworks:

Companies that successfully leverage NoSQL technologies are going to create massive value for themselves, their investors and their customers. But there’s no guarantee that this value creation will extend to independent, open source NoSQL vendors. It might, but make no mistake, building a successful open source NoSQL company is going to be a long, hard slog.

The same is true of Hadoop vendors or, more accurately stated, Hadoop vendors that plan to give away all their software in return for some small percentage of users turning into buyers. It can happen, but the odds are not in Hortonworks' favor. The company had a good quarter, but it will need many more to take its place next to Red Hat.

Related articles


Copyright © 2015 IDG Communications, Inc.